Home FruitVision: A deep learning based automatic fruit grading system
Article Open Access

FruitVision: A deep learning based automatic fruit grading system

  • Ahatsham Hayat EMAIL logo , Fernando Morgado-Dias , Tanupriya Choudhury , Thipendra P. Singh and Ketan Kotecha
Published/Copyright: May 15, 2024
Download Article (PDF)
Open Agriculture
From the journal Open Agriculture Volume 9 Issue 1

Abstract

Quality assessment of fruits plays a key part in the global economy’s agricultural sector. In recent years, it has been shown that fruits are affected by different diseases, which can lead to widespread economic failure in the agricultural industry. Traditional manual visual grading of fruits could be more accurate, making it difficult for agribusinesses to assess quality efficiently. Automatic grading of fruits using computer vision has become a prominent area of study for many researchers. In this study, a deep learning-based model called FruitVision is proposed for the automatic grading of various fruits. The results showed that FruitVision performed all the existing models and obtained an accuracy of 99.42, 99.50, 99.24, 99.12, 99.38, 99.38, 99.17, 98.86, and 97.96% for the apple, banana, guava, lime, orange, pomegranate, Ajwa date, Mabroom date, and mango, respectively, using 5-fold cross-validation. This is a remarkable achievement in the field of AI-based fruit grading systems.

Keywords: machine learning; computer vision; deep learning; fruit grading; agriculture

1 Introduction

Agriculture is vital to countries’ economies of most developing nations, including some highly populated countries like China and India [1]. Producing and distributing fresh fruits and vegetables to vendors and marketplaces worldwide is an essential aspect of agriculture. Several novel technologies have been created and employed in the food business to meet the rising demand for adequate food production and market supply [2]. Governments invest a significant amount of money each year in new technologies, and new agricultural practices and tactics are being deployed to combat pests, natural catastrophes, and drought [3]. Many merchants have increased their delivery of fruit in recent years. Many businesses seek high-quality food to compete for clients. As a result, retailers pressure suppliers to achieve quality standards [4].

Fruit harvesting involves examining the fruits and assessing the yield. Fruit grading and rating are difficult but necessary tasks since they help decide the fruit category and price. The automation of fruit grading has become a critical commercial operation due to the growing demand for high-quality fruits [5]. In many countries worldwide, fruit quality assessment primarily relies on human experts. Manual sorting through visual inspection is not only time-consuming and labor-intensive but also susceptible to human errors, given its manual nature. The aim of automating the grading process is to reduce labor costs while simultaneously enhancing efficiency and accuracy in the sorting process [6]. Because of the lack of knowledge among rural farmers, there is illiteracy in sophisticated agricultural mechanization. To meet demand, agriculture must step up and increase the significance of the impact element. The use of machine vision to manage, calibrate, and undertake ongoing study is critical in the agricultural business [7]. These systems analyze images of fruits and vegetables, allowing them to be recognized, graded, and divided. As the global population grows, so does the need for high-quality fruits. Many nation’s gross domestic product (GDP) is reliant on exports. The current technologies for quality detection, sorting, and dispensing have a poor yield, are costly, and are challenging to operate. Therefore, quick and efficient fruit sorting is required to decrease the labor concentration and increase the productivity of high-quality fruits [8]. An automated fruit grading system helps both consumers and farmers by ensuring that the market is always filled with high-quality products. Techniques such as machine vision and image processing have managed to gain popularity in the fruit industry, particularly in the areas of quality assurance and defect grading. Before being supplied to other marketplaces, the farm’s fruits are carefully inspected for quality [9].

Computer Vision and image processing became one of the most widely used artificial intelligence techniques for automatic fruit grading because of their high computational power and feasible results. Fruits can be categorized based on various components like size, shape, color, and texture. Image processing using machine learning techniques requires hand-craft feature extraction methods, which are time-consuming, and the same features cannot be used for every fruit. Alternatively, deep-learning techniques can be used as they extract relevant features automatically from the fruit image, and deep learning is a subset of machine learning that utilizes multilayered artificial neural networks to enable machines to classify images [10]. Therefore, in this study, a novel deep-learning architecture FruitVision was proposed for automatic fruit grading. The proposed architecture was based on MobileNet architecture [11], which was pre-trained on the ImageNet dataset. The performance of the proposed architecture was also compared with those of different state-of-the-art deep-learning architectures.

This study opens up new opportunities to use AI technology for sorting fruits on large-scale production lines where speed is an important factor to consider, along with the precision rate that this system offers over manual processing techniques like visual inspection or tactile sensing, etc., It also allows us to create better predictive models by analyzing data collected from different sources such as soil condition or climate conditions so that we can further improve our prediction accuracy while maintaining low-cost operation. This study is organized as follows; Section 2 focuses on recent studies on automatic fruit grading using machine learning and deep-learning techniques. Section 3 describes the different methodologies used in this study with evaluation criteria to check the performance of the methods. Section 4 discusses the experimental results of this study. Section 5 concludes the article with the pros and some limitations of this study.

2 Related work

Several studies have been conducted in the past for the automatic grading of fruits and vegetables. This literature review focused on two major techniques used for the automatic grading of fruits: machine learning-based and deep-learning-based.

Recent attempts have been made to use image processing and machine learning methods to automate the categorization of fruits based on their exterior appearance or freshness. Machine learning-based techniques usually try to predict fruit quality through hand-crafted features like shape, color texture, etc.

Bhargava et al. [12,13,14] use statistical, color, geometrical, and textural features to identify fruits and vegetables and grade them accordingly. In their research, a range of machine learning techniques were employed, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Among these methods, SVM has consistently demonstrated promising results across all aspects of their study.

Dubey and Jalal [15] and Singh and Singh [16] have used the texture, shape, and color features of apples and graded them into different categories. Dubey et al. [17] in their study got the best results after combining all the features, provided them to the multiclass SVM to grade the apples, and achieved an accuracy of 95.94%. Singh et al. [16] classified apples based on different features namely Gray Level Co-occurrence Matrix (GLCM), Discrete Wavelet Transform (DWT), Histogram of Oriented Gradients (HOG), and Tamura’s Features, Law’s Texture Energy. A total of 108 features are extracted and fed into various machine-learning models. Out of all types of models, SVM gives the best accuracy of 98.9%.

Furthermore, Moallem et al. [17] use multilayer perceptron for the segmentation of defective regions in apples and then features based on various textures, statistics, and geometry were extracted and fed to machine learning models. SVM classifier gives the best results with an accuracy of 92.5% for categorizing healthy and defective apples and 89.2% accuracy for three categories (First rank, Second rank, and Rejected).

Nandi et al. [18] and Pise et al. [19] in their study graded mango into different categories using machine learning methods. Nandi et al. [18] proposed an automated grading system for grading the mangos based on their maturity, fuzzy incremental learning algorithm was used to grade the mangos based on their maturity level, and the proposed system gave an accuracy of 87%.

Hashim et al. [20] proposed a system for banana quality estimation based on texture features. Classification was based done using ANN and SVM. Sabzi et al. [21] utilized an Artificial Neural Network and Particle Swarm Optimization algorithm (ANN-PSO) to automate the selection of the most influential features for grading oranges. Then, different classifiers were applied for classification comparison, out of which hybrid Artificial Neural Network–Artificial Bee Colony (ANN-ABC) gives the best accuracy of 96.70%.

Lara-Espinoza et al. [22] use ANN for grading guava fruit into three grades. They used three different color spaces RGB, CIELab, and CIELuv, for feature extraction and obtained an accuracy of 97.44%. Although machine learning‒based methods are showing good performance, they depend on hand-crafted features, and these are dependent on particular fruit. Therefore, developing a generalized system that can grade all the fruits can be a tedious task using machine learning techniques.

Deep learning-based techniques can be used as an alternative approach for the above problem for the fruit grading task. Ismail et al. [10,23], in their study, proposed a real-time visual inspection system for fruit grading using deep-learning techniques. They have used the banana and apple datasets for training and testing. They have used various state-of-the-art deep-learning methods to check the performance of the system. EfficientNet model gives the best accuracy of 99.2 and 98.60% for the apple and banana datasets, respectively. Similarly, Pande et al. [9] proposed a standalone system that can grade apples, oranges, and pears according to their maturity. They have used different pre-trained deep-learning models to classify fruits based on their quality. Among all fruits, the apple fruit was taken as a case study, applied the InceptionV3 model to grade the apple in four different grades, and achieved the best accuracy of 90%.

Nasiri et al. [24] and Raissouli et al. [25] used a deep CNN model for grading the dates. Raissouli et al. [25] achieved an incredible accuracy of 98% for the Ajwa date, 99% for the Mabroom date, and 99% for the Sukkary date. Vasumathi and Kamarasan [26] in their study proposed a deep CNN LSTM model to grade the pomegranate into normal and abnormal categories based on the features and visual features of the pomegranate.

Ucat and Cruz [27] and Jijesh et al. [28] proposed a Convolutional Neural Network (CNN) model to grade fruits. Ucat and Cruz [27] classify bananas into five different classes using CNN. First image thresholding was applied to get the segmented area, and then, it fed into CNN for classification, which gives an accuracy of 90%. Jijesh et al. [28] proposed a deep learning‒based fruit detection and grading system for apples. They have graded the apple into 3 different categories based on their quality. The accuracy of the system is also compared with some state-of-the-art methods. The proposed CNN outperformed other methods and gives an accuracy of 96.66%.

Xue et al.’s [29] study presents a hybrid deep-learning framework, CAE-ADN, for fruit image classification, which uses a convolution autoencoder for pre-training and an attention-based DenseNet for feature extraction. The framework has proven effective in improving fruit sorting efficiency, thereby reducing costs in the fresh supply chain.

Joseph et al.’s [30] study presents the development of a deep-learning model for fruit classification, which is crucial in various industries and dietary planning. The model, built with TensorFlow and trained on the Fruits 360 dataset, achieved an accuracy of 94.35% and offers a more efficient alternative to manual sorting methods.

It is observed from the above literature review that, many studies have been conducted in the past using machine learning-based and deep learning‒based techniques for fruit grading but only a few studies have focused on making a generalized system to grade multiple fruits, and some of the studies have a very small dataset, due to which, it is difficult to evaluate the system efficacy. Therefore, this study proposed to develop a generalized system for automatic fruit grading, which can be used for grading multiple fruits.

3 Methodology

The objective of this work is to develop a deep learning‒based automatic system for evaluating multiple fruits into different grades based on their quality. Figure 1 depicts the proposed methodology for an automated fruit grading system. This system is further benchmarked against a variety of state-of-the-art deep-learning techniques to assess its performance.

Figure 1 Workflow of proposed FruitVision system for fruit grading.
Figure 1

Workflow of proposed FruitVision system for fruit grading.

3.1 Dataset

The proposed system used three different databases to grade multiple fruits, i.e., apple, banana, guava, lime, orange, pomegranate, dates, and mango. The first dataset was FruitNet [31] which consists of six fruits namely apple, banana, guava, lime, orange, and pomegranate. The dataset is categorized into three categories (a. Good quality, b. Bad quality, and c. Mixed Quality). The dataset contains a total of 19526 images, and all of them were resized to 256 × 256-pixel sizes for further processing. A total number of Good, Bad, and Mixed quality images are 11,664, 6,778, and 1,074, respectively. Figure 2 shows some of the sample images of the FruitNet dataset.

Figure 2 Sample images from the FruitNet Dataset: (a) Good quality, (b) Bad quality, and (c) Mixed quality.
Figure 2

Sample images from the FruitNet Dataset: (a) Good quality, (b) Bad quality, and (c) Mixed quality.

The second dataset was the Taibah University-Dates Grading dataset (TU-DG) [25], which consists of three types of data categories, namely Ajwa, Mabroom, and Sukkary. In this study, Ajwa and Mabroom dates are taken for grading, both are categorized into three grades according to their quality which depends on size, weight, shape, etc. (grade 1 – Good quality, grade 2 – Average quality, and grade 3 – Bad quality). Total number of Grade 1, Grade 2, and Grade 3 images are 600, 845, and 988, respectively. The dataset contains a total of 2433 images of Ajwa and Mabroom, and Figure 3 shows some of the sample images of the TU-DG dataset.

Figure 3 Sample images from the TU-DG Dataset: (a) Grade 1, (b) Grade 2, and (c) Grade 3.
Figure 3

Sample images from the TU-DG Dataset: (a) Grade 1, (b) Grade 2, and (c) Grade 3.

The third dataset was the Mango Variety and Grading Dataset [32]. Mangos were graded into three different grades based on their quality (Extra Class – Good quality, Class 1 – Average quality, and Class 2 – Bad quality). The dataset contains a total of 600 images; Figure 4 shows some of the sample images of the mango dataset. Each class contains 200 images. Table 1 shows the important characteristics of all three datasets.

Figure 4 Sample images from the Mango Dataset: (a) Extra Class, (b) Class 1, and (c) Class 2.
Figure 4

Sample images from the Mango Dataset: (a) Extra Class, (b) Class 1, and (c) Class 2.

Table 1

General characteristics of the datasets

Sr. No. Fruit category Total no. of fruits Fruit categories
1 Apple 2,403 1. Good Quality, 2. Bad Quality, 3. Mixed Quality
2 Banana 2,485
3 Guava 2,429
4 Lime 2,457
5 Orange 2,500
6 Pomegranate 7,252
7 Ajwa Date 1,137 1. Grade 1, 2. Grade 2, 3. Grade 3
8 Mabroom Date 1,296
9 Mango 600 1. Extra Class, 2. Class 1, 3. Class 2

3.2 Image pre-processing

3.2.1 Contrast limited adaptive histogram equalization (CLAHE)

Images obtained using diverse approaches include many types of noise that decrease image quality and do not give accurate information for later processing. Therefore, to eliminate any undistributed illumination and noise in images, CLAHE [33] was applied to the dataset which is a local histogram equalization method. CLAHE algorithm will divide the image into several non-overlapping regions called tiles, and to eliminate the artificial boundaries, the adjacent tiles are merged using a bilinear interpolation function. In this study, the CLAHE algorithm is applied to the color image with a tile size of 8 × 8 and a clip limit of 20. Figure 5 shows the histogram of the original and CLAHE processed image.

Figure 5 Image enhancement using CLAHE on Mango Dataset.
Figure 5

Image enhancement using CLAHE on Mango Dataset.

From Figure 5, it is possible to say that after applying the CLAHE enhancement algorithm, images are smoother, and illumination is distributed uniformly.

3.2.2 Segmentation

Since all of the images present in the datasets have a different background, it is important to separate the fruit image from the background. In this study, Otsu’s thresholding [34] technique was used for the background segmentation. Otsu’s thresholding is basically based on the histogram method, and the main idea is to separate the fruit image into two clusters using a threshold determined by minimizing the weighted variance of these classes. Equation (1) shows the mathematical computation of Otsu’s thresholding.

(1) σ o 2 ( t ) = w 1 ( t ) σ 1 2 ( t ) + w 2 ( t ) σ 2 2 ( t ) ,

where w 1(t) and w 2(t) represent the probabilities of two classes divided by the threshold t.

Figure 6 shows the overall image preprocessing results after the image enhancement and segmentation.

Figure 6 Overall preprocessing results.
Figure 6

Overall preprocessing results.

3.3 Model architecture

FruitVision is a deep CNN architecture tailored for grading various fruits. It is based on the MobileNetV3 architecture [11]. MobileNetV3 uses two techniques called MnasNet [35] and NetAdapt [36] in sequence to exploit the search space in order to get the optimized architecture. MnasNet [35] selects the optimal configuration for the network and then the network is fine-tuned using NetAdapt [36]. Additionally, a new activation function was introduced in MobileNetV3 called hard-swish (h-swish), which is a nonlinear function based on the Swish function [11]. Equation (2) shows the mathematical representation of h-swish function. The general swish function uses the sigmoid function which requires high computational resources on mobile devices. To solve this problem, h-swish was introduced which uses the rectified linear unit activation function6(ReLU6) function instead of sigmoid.

(2) h - swish ( x ) = x · ReLU 6 ( x + 3 ) 6 .

As shown in Figure 7, the backbone of the proposed architecture has an inverted residual block. The Inverted Residual block in this architecture contains a depthwise separable convolutional block as an efficient replacement of the conventional convolutional block, squeeze, and excitement block [37], which assigns greater importance to the pertinent features present in each channel. The Depthwise Separable Convolutional Block is comprised of two distinct layers: a lightweight depthwise convolutional kernel that operates on each channel to perform spatial filtering and a more substantial 1 × 1 pointwise convolutional kernel accompanied by a batch normalization (BN) layer. This is followed by the application of the h-swish function to generate features. After the feature extraction process, a flattening layer was applied, followed by the two fully connected dense layers at the end. FruitVision has 6,007,171 total parameters of which 3,010,819 are trainable, and 2,996,352 are non-trainable parameters. Table 2 shows the architectural details of the FruitVision model.

Figure 7 Architecture of the FruitVision Model.
Figure 7

Architecture of the FruitVision Model.

Table 2

Details of the FruitVision architecture

Layer name Output shape Parameters
MobileNetV3 (Model) (7, 7, 960) 0
flatten (Flatten) 47040 0
dense_1 (Dense) 64 3010624
dropout (Dropout) 64 0
dense_2 (Dense) 3 195
Total params: 6,007,171
Trainable params: 3,010,819
Non-trainable params: 2,996,352

3.4 Model training

The proposed FruitVision model was pre-trained on ImageNet dataset and then fine-tuned on the different datasets discussed in this study using a transfer learning approach. Adam optimizer with the default learning rate of 0.001 was used in the training process with a batch size of 16. All the datasets were divided randomly into three sets (training, testing, and validation) with a split ratio of 70, 15, and 15% for training, testing, and validation sets, respectively. A 5-fold cross-validation was used to assess the performance of the models and model training was set for 100 epochs cycle with the stopping criteria (10 patience). Finally, the fine-tuned proposed network was tested on the test dataset.

3.5 Evaluation criteria

To evaluate the performance of the developed models was based on the standard measures that use the True Positive (TP), True Negative (TN), False Positive (FP), and False Negative (FN). Equations (3)–(7) show the mathematical form of these standard measures.

(3) Accuracy = TP + TN TP + TN + FP + FN ,

(4) Specificity = TN TP + FP ,

(5) Precision = TP TP + FP ,

(6) Recall = TP TP + FN ,

(7) F 1 Score = 2 × Precision × Recall Precision + Recall .

The performance of the proposed FruitVision model was also compared with various pre-trained deep-learning CNN models. Section 4 discussed the results obtained from the FruitVision and other deep-learning models.

To check the statistical significance difference in model performance, we have performed a one-way ANOVA test [38] which will tell if the models are significantly different or not in the mean accuracies of the models. Furthermore, to investigate individual pairs of differences in models, we conducted a posthoc Tukey HSD test [39], and this test will reveal the significant difference between pairs of models.

4 Results and discussions

Fruit quality estimation and proper grading are very important tasks in the agriculture field. Automatic fruit grading can help farmers and retailers to get the maximum price for their fruits. The quality of the extracted features will determine the performance of any deep-learning model; therefore, extracting features from fruit images is an important step in training models. Figure 8 shows some features generated by the first convolutional layer of the FruitVision model for the banana dataset.

Figure 8 Deep Features learned by FruitVision Model for the Banana Dataset.
Figure 8

Deep Features learned by FruitVision Model for the Banana Dataset.

The accomplishments of deep-learning models in the field of computer vision, served as the motivation for this study, which has the purpose to conduct a comprehensive study on the grading of fruit images. To compare the performance of the proposed FruitVision model, several state-of-the-art deep-learning models are also implemented. For better-generalized performance of each model, the 5-fold cross-validation technique was used.

Tables 3–8 show the overall performance comparison of all pre-trained models, namely, VGG19 [40], ResNet 50 [41], ResNet101 [41], DenseNet121 [42], DenseNet201 [42], MobileNetV3 [11], InceptionV3 [43], and NASNetMobile [44] with the proposed FruitVision model for the first dataset. In Tables 3–8, the best results are highlighted for easy understanding. The “Group” column in the table shows the statistical grouping of the models based on the Tukey HSD post-hoc test. Models that have the same letter do not have significant differences in their performance statistically. For example, in Table 3, DenseNet101 and DenseNet201 (Group Letter “D”) do not have a statistical difference in their performance.

Table 3

Overall Performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Apple dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 94.23 ± 0.02 95.20 ± 0.01 94.55 ± 0.03 94.87 ± 0.04 93.41 ± 0.68 A
ResNet50 95.78 ± 0.07 96.02 ± 0.05 95.27 ± 0.29 95.64 ± 0.32 94.28 ± 0.37 B
ResNet101 96.41 ± 0.10 96.33 ± 0.33 95.64 ± 0.65 95.98 ± 0.12 94.84 ± 0.22 C
DenseNet121 98.54 ± 0.22 97.48 ± 0.02 96.21 ± 0.38 96.84 ± 0.20 96.34 ± 0.38 D
DenseNet201 99.10 ± 0.17 97.89 ± 0.47 97.14 ± 0.52 97.51 ± 0.64 96.89 ± 0.10 D
MobileNetV3 96.47 ± 0.11 96.92 ± 0.35 95.76 ± 0.66 96.34 ± 0.28 95.64 ± 0.75 E
InceptionV3 92.10 ± 0.21 93.74 ± 0.17 93.27 ± 0.50 93.50 ± 0.32 92.51 ± 0.68 F
NASNetMobile 95.48 ± 0.31 96.67 ± 0.64 95.63 ± 0.31 96.15 ± 0.74 93.47 ± 0.34 G
FruitVision (Proposed) 99.42 ± 0.10 99.58 ± 0.37 98.71 ± 0.25 99.14 ± 0.20 97.34 ± 0.54 H
Table 4

Overall Performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Banana dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 97.54 ± 0.17 97.25 ± 0.34 96.85 ± 0.21 97.05 ± 0.48 96.12 ± 0.21 A
ResNet50 96.21 ± 0.62 96.42 ± 0.27 96.05 ± 0.68 96.23 ± 0.30 95.74 ± 0.38 B
ResNet101 98.30 ± 0.04 97.66 ± 0.31 96.94 ± 0.37 97.30 ± 0.49 95.76 ± 0.62 A
DenseNet121 98.42 ± 0.33 97.21 ± 0.28 97.14 ± 0.24 97.17 ± 0.33 96.27 ± 0.84 A
DenseNet201 98.84 ± 0.28 98.35 ± 0.45 97.51 ± 0.29 97.93 ± 0.67 97.10 ± 0.36 A
MobileNetV3 97.24 ± 0.43 96.82 ± 0.36 96.83 ± 0.47 96.82 ± 0.31 96.88 ± 0.22 C
InceptionV3 94.11 ± 0.57 94.21 ± 0.65 93.64 ± 0.62 93.92 ± 0.57 93.27 ± 0.26 D
NASNetMobile 96.74 ± 0.27 96.72 ± 0.38 96.25 ± 0.34 96.48 ± 0.63 96.38 ± 0.65 E
FruitVision (Proposed) 99.50 ± 0.20 99.19 ± 0.28 98.88 ± 0.74 99.03 ± 0.55 98.77 ± 0.34 A
Table 5

Overall Performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Guava dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 95.21 ± 0.15 95.74 ± 0.44 95.15 ± 0.60 95.44 ± 0.38 94.57 ± 0.32 A
ResNet50 96.00 ± 0.74 96.06 ± 0.61 95.34 ± 0.22 95.70 ± 0.70 95.62 ± 0.11 B
ResNet101 96.52 ± 0.21 96.28 ± 0.37 96.21 ± 0.30 96.24 ± 0.65 95.71 ± 0.24 C
DenseNet121 97.77 ± 0.64 97.87 ± 0.48 96.85 ± 0.62 97.36 ± 0.22 96.32 ± 0.18 C
DenseNet201 98.94 ± 0.07 98.65 ± 0.14 97.78 ± 0.52 98.21 ± 0.16 97.12 ± 0.37 C
MobileNetV3 96.70 ± 0.34 97.11 ± 0.21 96.22 ± 0.49 96.66 ± 0.72 96.23 ± 0.28 D
InceptionV3 95.74 ± 0.74 95.37 ± 0.26 95.13 ± 0.28 95.25 ± 0.25 94.86 ± 0.46 E
NASNetMobile 96.40 ± 0.31 96.80 ± 0.41 96.32 ± 0.35 96.56 ± 0.40 95.93 ± 0.30 C
FruitVision (Proposed) 99.24 ± 0.27 99.32 ± 0.32 98.85 ± 0.47 99.08 ± 0.37 98.48 ± 0.15 C
Table 6

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Lime dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 95.35 ± 0.42 95.21 ± 0.31 94.95 ± 0.14 95.08 ± 0.74 94.82 ± 0.49 A
ResNet50 95.28 ± 0.37 95.30 ± 0.25 95.11 ± 0.27 95.20 ± 0.29 95.40 ± 0.37 B
ResNet101 96.87 ± 0.24 96.42 ± 0.18 96.50 ± 0.56 96.46 ± 0.45 96.75 ± 0.68 C
DenseNet121 97.90 ± 0.57 97.58 ± 0.34 98.10 ± 0.24 97.84 ± 0.68 97.05 ± 0.25 C
DenseNet201 98.48 ± 0.30 98.10 ± 0.42 98.76 ± 0.34 98.43 ± 0.72 97.90 ± 0.10 C
MobileNetV3 97.67 ± 0.64 97.61 ± 0.72 98.05 ± 0.47 97.83 ± 0.38 97.22 ± 0.36 C
InceptionV3 94.20 ± 0.21 94.05 ± 0.37 94.66 ± 0.65 94.35 ± 0.26 93.65 ± 0.27 D
NASNetMobile 97.24 ± 0.43 97.14 ± 0.45 97.20 ± 0.35 97.17 ± 0.15 96.52 ± 0.68 C
FruitVision (Proposed) 99.12 ± 0.12 98.76 ± 0.28 99.25 ± 0.21 99.00 ± 0.18 98.88 ± 0.35 C
Table 7

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Orange dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 96.15 ± 0.25 96.42 ± 0.38 95.65 ± 0.33 96.03 ± 0.27 95.82 ± 0.54 A
ResNet50 96.82 ± 0.34 96.94 ± 0.14 96.78 ± 0.52 96.86 ± 0.38 96.21 ± 0.26 B
ResNet101 97.12 ± 0.19 97.08 ± 0.90 97.20 ± 0.69 97.14 ± 0.64 97.10 ± 0.37 C
DenseNet121 98.03 ± 0.74 98.15 ± 0.29 97.33 ± 0.44 97.74 ± 0.52 97.95 ± 0.16 B
DenseNet201 98.74 ± 0.65 98.58 ± 0.86 98.49 ± 0.39 98.53 ± 0.37 98.35 ± 0.08 B
MobileNetV3 97.57 ± 0.38 98.05 ± 0.62 96.82 ± 0.61 97.43 ± 0.45 97.21 ± 0.21 D
InceptionV3 95.31 ± 0.28 95.45 ± 0.39 94.27 ± 0.33 94.86 ± 0.64 95.12 ± 0.36 E
NASNetMobile 97.17 ± 0.18 97.55 ± 0.46 96.41 ± 0.24 96.98 ± 0.50 96.90 ± 0.12 F
FruitVision (Proposed) 99.38 ± 0.12 99.40 ± 0.29 99.02 ± 0.22 99.21 ± 0.37 98.83 ± 0.40 B
Table 8

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Pomegranate dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 96.07 ± 0.47 94.55 ± 0.67 96.65 ± 0.31 95.59 ± 0.09 94.21 ± 0.22 A
ResNet50 97.52 ± 0.36 94.92 ± 0.22 97.70 ± 0.55 96.29 ± 0.38 95.12 ± 0.31 B
ResNet101 97.85 ± 0.84 95.64 ± 0.37 97.88 ± 0.25 96.75 ± 0.25 95.25 ± 0.19 C
DenseNet121 98.64 ± 0.23 96.14 ± 0.41 98.47 ± 0.30 97.29 ± 0.12 96.34 ± 0.27 D
DenseNet201 98.90 ± 0.44 96.59 ± 0.16 99.12 ± 0.41 97.84 ± 0.37 96.42 ± 0.16 D
MobileNetV3 98.22 ± 0.30 95.37 ± 0.27 98.35 ± 0.67 96.84 ± 0.85 95.65 ± 0.34 E
InceptionV3 94.98 ± 0.56 92.28 ± 0.32 95.37 ± 0.50 93.80 ± 0.41 92.54 ± 0.28 F
NASNetMobile 98.27 ± 0.43 95.68 ± 0.52 98.52 ± 0.47 97.08 ± 0.38 95.40 ± 0.44 G
FruitVision (Proposed) 99.31 ± 0.11 97.52 ± 0.58 99.60 ± 0.20 98.55 ± 0.44 97.87 ± 0.08 D

The proposed FruitVision model achieved a remarkable accuracy of 99.42, 99.50, 99.24, 99.12, 99.38, and 99.31% for the apple, banana, guava, lime, orange, and Pomegranate, respectively. The second-best accuracy for all the fruits present in the first dataset was obtained by the DanseNet201 model. Figure 9 shows the training and validation loss curve for all the fruits present in the first dataset. For all the fruits, the model was stopped with the early stopping criteria, which means that the model achieved the optimal loss. Furthermore, from Figure 9, it is observed that for all the fruits in the first dataset training and validation loss of the model aligned with each other at the end, and this implies that the proposed model is tuning well during the training process.

Figure 9 Training and validation loss curve for the First Dataset fruits: (a) apple, (b) banana, (c) guava, (d) lime, (e) orange, and (f) pomegranate.
Figure 9

Training and validation loss curve for the First Dataset fruits: (a) apple, (b) banana, (c) guava, (d) lime, (e) orange, and (f) pomegranate.

Proposed FruitVision shows higher and more consistent accuracy in all the datasets, showing the robustness of the model. Despite the fact that in some cases, the proposed model is statistically similar to DenseNet-based models, and its performance across various datasets attests to the generalizability of the proposed model.

After analyzing Tables 3–8, it is possible to say that the proposed FruitVision architecture outperformed all the other pre-trained models for the first dataset.

Now, it is possible to analyze the performance of the proposed methods for the second dataset. Tables 9 and 10 show the overall performance comparison between the proposed and pre-trained neural networks. For the second dataset, the proposed system gives the best accuracy of 99.17, and 98.86% for Ajwa and Mabroom dates. Figure 10 shows the training and validation loss curve for the second dataset.

Table 9

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Ajwa Date dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 95.77 ± 0.64 97.24 ± 0.72 95.33 ± 0.20 96.28 ± 0.78 94.75 ± 0.32 A
ResNet50 97.44 ± 0.30 96.52 ± 0.38 96.76 ± 0.39 96.64 ± 0.28 93.97 ± 0.37 B
ResNet101 98.14 ± 0.09 97.37 ± 0.27 97.64 ± 0.42 97.50 ± 0.14 95.54 ± 0.11 C
DenseNet121 98.58 ± 0.30 97.14 ± 0.45 97.93 ± 0.63 97.53 ± 0.30 97.39 ± 0.68 D
DenseNet201 99.11 ± 0.12 98.65 ± 0.38 98.40 ± 0.14 98.52 ± 0.46 98.20 ± 0.42 D
MobileNetV3 98.11 ± 0.33 98.22 ± 0.28 97.14 ± 0.27 97.68 ± 0.55 97.75 ± 0.29 E
InceptionV3 96.37 ± 0.76 96.95 ± 0.56 95.76 ± 0.35 96.35 ± 0.25 94.97 ± 0.37 F
NASNetMobile 98.31 ± 0.25 98.30 ± 0.14 98.05 ± 0.64 98.17 ± 0.74 97.54 ± 0.28 E
FruitVision (Proposed) 99.17 ± 0.31 99.50 ± 0.25 98.65 ± 0.29 99.07 ± 0.37 98.83 ± 0.21 D
Table 10

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Mabroom Date dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 97.65 ± 0.28 97.41 ± 0.21 96.62 ± 0.31 97.01 ± 0.63 96.75 ± 0.22 A
ResNet50 96.23 ± 0.39 96.65 ± 0.28 95.77 ± 0.27 96.21 ± 0.24 95.34 ± 0.38 A
ResNet101 96.30 ± 0.24 96.92 ± 0.75 95.85 ± 0.38 96.38 ± 0.29 96.14 ± 0.40 B
DenseNet121 97.11 ± 0.56 98.03 ± 0.39 96.64 ± 0.64 97.33 ± 0.25 96.82 ± 0.28 C
DenseNet201 97.63 ± 0.20 98.64 ± 0.25 97.21 ± 0.20 97.92 ± 0.45 96.91 ± 0.54 C
MobileNetV3 95.82 ± 0.71 96.42 ± 0.87 95.10 ± 0.12 95.76 ± 0.68 96.74 ± 0.38 D
InceptionV3 94.59 ± 0.68 95.35 ± 0.22 94.21 ± 0.37 94.78 ± 0.12 93.66 ± 0.37 E
NASNetMobile 95.77 ± 0.33 96.74 ± 0.14 95.62 ± 0.26 96.18 ± 0.14 96.10 ± 0.26 F
FruitVision (Proposed) 98.86 ± 0.26 98.96 ± 0.37 98.41 ± 0.31 98.68 ± 0.37 98.55 ± 0.28 C
Figure 10 Training and validation loss curve for the Second Dataset fruits: (a) Ajwa Date, and (b) Mabroom Date.
Figure 10

Training and validation loss curve for the Second Dataset fruits: (a) Ajwa Date, and (b) Mabroom Date.

After analyzing Figure 10, it is possible to say that the proposed FruitVision architecture achieved the optimal validation loss as it is stopped with the Early Stopping criteria. Furthermore, in the beginning, the gap between the training and validation loss was very high but in the end training and validation loss was almost converging, which shows that the proposed model fits very well the dataset.

Table 11 shows the overall performance comparison between the proposed and various state-of-art deep-learning models for the third dataset (Mango Dataset). The proposed architecture gives the best accuracy of 97.96%, precision of 97.21%, recall of 97.10%, f1-score of 97.15%, and specificity of 95.87% among all. DenseNet201 shows the best results among all the existing pre-trained models and InceptionV3 obtained the poorest results.

Table 11

Overall performance comparison of proposed FruitVision model with other state-of-the-art deep-learning model for Mango Date dataset

Model Accuracy (%) Precision (%) Recall (%) F1 Score (%) Specificity (%) Group
VGG19 92.48 ± 0.30 93.34 ± 0.54 92.87 ± 0.64 93.10 ± 0.61 90.24 ± 0.43 A
ResNet50 94.16 ± 0.22 93.52 ± 0.35 93.70 ± 0.37 93.61 ± 0.26 92.77 ± 0.18 B
ResNet101 95.63 ± 0.44 93.85 ± 0.27 95.10 ± 0.28 94.47 ± 0.72 93.87 ± 0.26 B
DenseNet121 95.25 ± 0.25 95.66 ± 0.47 95.41 ± 0.61 95.53 ± 0.27 93.63 ± 0.60 C
DenseNet201 96.71 ± 0.37 96.78 ± 0.38 96.46 ± 0.27 96.62 ± 0.38 94.10 ± 0.28 D
MobileNetV3 96.12 ± 0.45 95.55 ± 0.64 95.75 ± 0.29 95.65 ± 0.62 93.86 ± 0.16 E
InceptionV3 91.87 ± 0.81 92.45 ± 0.25 90.34 ± 0.34 91.38 ± 0.24 89.74 ± 0.42 F
NASNetMobile 94.52 ± 0.49 95.88 ± 0.27 93.44 ± 0.51 94.64 ± 0.10 92.33 ± 0.25 G
FruitVision (Proposed) 97.96 ± 0.24 97.21 ± 0.37 97.10 ± 0.62 97.15 ± 0.52 95.87 ± 0.32 D

Figure 11 shows the training and validation loss of the proposed FruitVision model for the mango dataset. Initially, the training loss was very high, but after a few epoch cycles, both the training and validation loss were almost converging, and at the end, both the losses were less than 0.20, which demonstrates how well the model fits the training and validation set.

Figure 11 Training and validation loss curve for the third dataset (Mango Dataset).
Figure 11

Training and validation loss curve for the third dataset (Mango Dataset).

Figures 12–14 show the prediction results of the proposed FruitVision model for all the datasets used in this study. After reviewing Tables 3–11 and Figures 12–14, it is possible to say that the proposed model gives satisfactory performance for all the dataset and can be used in the agriculture industry for automizing the process of fruit grading which will enhance the accuracy and speed up the process of grading the fruits.

Figure 12 Prediction results of Proposed FruitVision model for the First Dataset.
Figure 12

Prediction results of Proposed FruitVision model for the First Dataset.

Figure 13 Prediction results of the proposed FruitVision model for the Second Dataset.
Figure 13

Prediction results of the proposed FruitVision model for the Second Dataset.

Figure 14 Prediction results of Proposed FruitVision model for the Third Dataset.
Figure 14

Prediction results of Proposed FruitVision model for the Third Dataset.

The proposed solution was also compared with the recent studies which use machine and deep-learning methods for the automatic grading of fruits. In machine-based methods, SVM had obtained the most satisfactory results among all the other methods. In the deep learning-based method, the proposed FruitVision model outperformed all of the recent studies done for the fruit grading. Table 12 shows the comparative analysis of the proposed method with the existing studies.

Table 12

Comparative analysis of the proposed method with the existing research for fruit grading

Author Fruits Best Model Accuracy (%)
Bhargava et al. [12] Apple, avocado, banana, and orange SVM 96.59
Pande et al. [9] Apple Inception V3 90
Ismail et al. [23] Apple, banana EfficientNet Apple – 99.20
Banana – 98.50
Bhargava et al. [12] Apple, avocado, banana, and orange SVM 98.48
Bhargava et al. [13] Apple SVM 98.42
Dubey and Jalal [15] Apple MSVM 95.94
Singh et al. [16] Apple SVM 98.90
Moallem et al. [17] Apple SVM 92.50
Nandi et al. [18] Mango Fuzzy Incremental Learning 87
Sabzi et al. [21] Orange ANN-ABC 96.70
Lara-Espinoza et al. [22] Guava ANN 97.44
Nasiri et al. [24] Date CNN 96.98
Raissouli et al. [25] Date CNN Ajwa Date – 98
Mabroom – 99
Vasumathi et al. [26] Pomegranate CNN LSTM 98.17
Ucat et al. [27] Banana CNN 90
Joseph et al. [30] Apple CNN 96.66
Proposed CNN Apple, banana, guava, lime, orange, pomegranate, date, and mango FruitVision (MobileNetV3 + CNN) Apple – 99.42
Banana – 99.50
Guava – 99.24
Lime – 99.12
Orange – 99.38
Pomegranate – 99.31
Ajwa Date – 99.17
Mabroom Date – 98.86
Mango – 97.96

5 Conclusion and future scope

Fruits are rich in nutrients, including oxidants and flavonoids, which reduce the incidence of heart disease, cancer, and high blood pressure. Defective fruits cause economic loss and may affect health. Identifying and evaluating fruit quality became a significant issue of research. Manual grading of fruits is time-consuming and error-prone. Therefore, it is important to have a computer vision-based system that will automatically grade the fruits based on their quality.

This article proposed a deep learning‒based computer vision model called FruitVision for the automatic grading of various fruits from their outer appearance. Various state-of-the-art deep-learning models were also implemented in this study to check the effectiveness of the proposed model. The proposed model outperformed all the existing models and obtained an accuracy of 99.42, 99.50, 99.24, 99.12, 99.38, 99.17, 98.86, and 97.96% for the apple, banana, guava, lime, orange, pomegranate, Ajwa date, Mabroom date, and mango fruit, respectively. Therefore, the proposed model can be used in the agriculture industry for grading fruits automatically.

The proposed deep-learning model has produced remarkable results, but it also has certain limitations. The first one is that the proposed model should be tested with larger and multiple fruit datasets for better generalization and robustness. The second limitation is that the proposed system was trained and tested on the single view of fruit images. It will be interesting to see the performance of the system trained on multi-view fruit images.

  1. Funding information: We thank to receive the funding for this from RSF (Research Support funding) from Symbiosis Institute of Technology, Symbiosis International (Deemed University) (SIU),Lavale Campus, Pune, Maharashtra, 412115, India. F. Morgado-Dias received funding by 10.54499/LA/P/0083/2020; 10.54499/UIDP/50009/2020 & 10.54499/UIDB/50009/2020.

  2. Author contributions: All authors accepted the responsibility for the content of the manuscript and consented to its submission, reviewed all the results, and approved the final version of the manuscript. AH and TC were responsible for investigation, original draft writing, and draft revisions, including methodology. FM-D, TPS, and KK provided supervision, conducted review and validation, and contributed to editing.

  3. Conflict of interest: Authors state no conflict of interest.

  4. Data availability statement: The dataset used in this article is openly available and taken from the FruitNet [31], (TU-DG) [25], and Mango Variety and Grading Dataset [32] papers.

References

[1] Arakeri MP, Lakshmana. Computer vision based fruit grading system for quality evaluation of tomato in agriculture industry. Procedia Comput Sci. 2016;79:426–33. 10.1016/j.procs.2016.03.055.Search in Google Scholar

[2] Tripathi MK, Maktedar DD. A role of computer vision in fruits and vegetables among various horticulture products of agriculture fields: A survey. Inf Process Agric. Jun. 2020;7(2):183–203. 10.1016/J.INPA.2019.07.003.Search in Google Scholar

[3] Huddar SR, Gowri S, Keerthana K, Vasanthi S, Rupanagudi SR. Novel algorithm for segmentation and automatic identification of pests on plants using image processing. 2012 3rd Int. Conf. Comput. Commun. Netw. Technol. ICCCNT 2012; 2012. 10.1109/ICCCNT.2012.6396012.Search in Google Scholar

[4] Clement J, Novas N, Gazquez JA, Manzano-Agugliaro F. High speed intelligent classifier of tomatoes by colour, size and weigh. Span J Agric Res. 2012;10(2):314–25. 10.5424/sjar/2012102-368-11.Search in Google Scholar

[5] Wu D, Sun DW. Colour measurements by computer vision for food quality control - A review. Trends Food Sci Technol. Jan. 2013;29(1):5–20. 10.1016/J.TIFS.2012.08.004.Search in Google Scholar

[6] Haggag M, Abdelhay S, Mecheter A, Gowid S, Musharavati F, Ghani S. An intelligent hybrid experimental-based deep learning algorithm for tomato-sorting controllers. IEEE Access. 2019;7:106890–98. 10.1109/ACCESS.2019.2932730.Search in Google Scholar

[7] Kamilaris A, Prenafeta-Boldú FX. Deep learning in agriculture: A survey. Comput Electron Agric. Apr. 2018;147:70–90. 10.1016/J.COMPAG.2018.02.016.Search in Google Scholar

[8] Mazen FMA, Nashat AA. Ripeness classification of bananas using an artificial neural network. Arab J Sci Eng. Jan. 2019;44(8):6901–10. 10.1007/S13369-018-03695-5.Search in Google Scholar

[9] Pande A, Munot M, Sreeemathy R, Bakare RV. An efficient approach to fruit classification and grading using deep convolutional neural network. 2019 IEEE 5th Int. Conf. Converg. Technol. I2CT 2019; Mar. 2019. 10.1109/I2CT45611.2019.9033957.Search in Google Scholar

[10] Baglat P, Hayat A, Mendonça F, Gupta A, Mostafa SS, Morgado-Dias F. Non-destructive banana ripeness detection using shallow and deep learning: A systematic review. Sensors. 2023;23(2):738. 10.3390/s23020738.Search in Google Scholar PubMed PubMed Central

[11] Howard A, et al. Searching for MobileNetV3. 2019 IEEE/CVF Int. Conf. Comput. Vis.; Oct. 2019. p. 1314–24. 10.1109/ICCV.2019.00140.Search in Google Scholar

[12] Bhargava A, Bansal A. Automatic detection and grading of multiple fruits by machine learning. Food Anal Methods. Mar. 2020;13(3):751–61. 10.1007/S12161-019-01690-6/TABLES/6.Search in Google Scholar

[13] Bhargava A, Bansal A. Classification and grading of multiple varieties of apple fruit. Food Anal Methods. 2021;1359–68. 10.1007/s12161-021-01970-0.Search in Google Scholar

[14] Bhargava A, Bansal A, Goyal V. Machine learning-based detection and sorting of multiple vegetables and fruits. Food Anal Methods. 2022;15(1):228–42. 10.1007/s12161-021-02086-1.Search in Google Scholar

[15] Dubey SR, Jalal AS. Apple disease classification using color, texture and shape features from images. Signal Image Video Process. 2016;10:819–26. 10.1007/s11760-015-0821-1.Search in Google Scholar

[16] Singh S, Singh NP. Machine learning-based classification of good and rotten apple. Vol. 524. Singapore: Springer; 2019. 10.1007/978-981-13-2685-1_36.Search in Google Scholar

[17] Moallem P, Serajoddin A, Pourghassem H. Computer vision-based apple grading for golden delicious apples based on surface features. Inf Process Agric. Mar. 2017;4(1):33–40. 10.1016/J.INPA.2016.10.003.Search in Google Scholar

[18] Nandi CS, Tudu B, Koley C. A machine vision technique for grading of harvested mangoes based on maturity and quality. IEEE Sens J. Aug. 2016;16(16):6387–96. 10.1109/JSEN.2016.2580221.Search in Google Scholar

[19] Pise D, Upadhye GD. Grading of harvested mangoes quality and maturity based on machine learning techniques. 2018 Int. Conf. Smart City Emerg. Technol. ICSCET 2018; Nov. 2018. 10.1109/ICSCET.2018.8537342.Search in Google Scholar

[20] Hashim N, Adebayo SE, Abdan K, Hanafi M. Comparative study of transform-based image texture analysis for the evaluation of banana quality using an optical backscattering system. Postharvest Biol Technol. Jan. 2018;135:38–50. 10.1016/J.POSTHARVBIO.2017.08.021.Search in Google Scholar

[21] Sabzi S, Abbaspour-Gilandeh Y, García-Mateos G. A new approach for visual identification of orange varieties using neural networks and metaheuristic algorithms. Inf Process Agric. Mar. 2018;5(1):162–72. 10.1016/J.INPA.2017.09.002.Search in Google Scholar

[22] Lara-Espinoza EM, Trejo-Duran M, Lizarraga-Morales RA, Cabal-Yepez E, Saldana-Robles N. Determination of the ripeness state of guavas using an artificial neural network. Res Comput Sci. Dec. 2016;121(1):105–11. 10.13053/RCS-121-1-9.Search in Google Scholar

[23] Ismail N, Malik OA. Real-time visual inspection system for grading fruits using computer vision and deep learning techniques. Inf. Process. Agric. 2021;9(1):24–37.10.1016/j.inpa.2021.01.005Search in Google Scholar

[24] Nasiri A, Taheri-Garavand A, Zhang YD. Image-based deep learning automated sorting of date fruit. Postharvest Biol Technol. Jul. 2019;153:133–41. 10.1016/J.POSTHARVBIO.2019.04.003.Search in Google Scholar

[25] Raissouli H, Aljabri AA, Aljudaibi SM, Haron F, Alharbi G. Date grading using machine learning techniques on a novel dataset. Int J Adv Comput Sci Appl. 2020;11(8):758–65. 10.21227/qhrr-m850.Search in Google Scholar

[26] Vasumathi MT, Kamarasan M. An effective pomegranate fruit classification based on CNN-LSTM deep learning models. Indian J Sci Technol. 2021;14(16):1310–9. 10.17485/IJST/v14i16.432.Search in Google Scholar

[27] Ucat RC, Cruz JCD. Postharvest grading classification of cavendish banana using deep learning and tensorflow. 2019 Int. Symp. Multimed. Commun. Technol. ISMAC 2019; Aug. 2019:758–65. 10.1109/ISMAC.2019.8836129.Search in Google Scholar

[28] Jijesh JJ, Shankar S, Ranjitha, Revathi DC, Shivaranjini M, Sirisha R. Development of machine learning based fruit detection and grading system. Proc - 5th IEEE Int Conf Recent Trends Electron Inf Commun Technol RTEICT 2020; Nov. 2020. p. 403–7. 10.1109/RTEICT49044.2020.9315601.Search in Google Scholar

[29] Xue G, Liu S, Ma Y. A hybrid deep learning-based fruit classification using attention model and convolution autoencoder. Complex Intell Syst. 2023;9:2209–19. 10.1007/s40747-020-00192-x.Search in Google Scholar

[30] Joseph JL, Kumar VA, Mathew SP. Fruit classification using deep learning. In: Mekhilef S, Favorskaya M, Pandey RK, Shaw RN, (eds). Innovations in electrical and electronic engineering. Lecture notes in electrical engineering. Vol. 756. Singapore: Springer; 2021. 10.1007/978-981-16-0749-3_62.Search in Google Scholar

[31] Meshram V, Patil K. FruitNet: Indian fruits image dataset with quality for machine learning applications. Data Brief. Feb. 2022;40:107686. 10.1016/J.DIB.2021.107686.Search in Google Scholar

[32] Rizwan Iqbal HM, Hakim A. Mango variety and grading dataset. Mendeley Data, V1. 2021;1. 10.17632/5MC3S86982.1.Search in Google Scholar

[33] Pizer SM, Amburn EP, Austin JD, Cromartie R, Geselowitz A, Greer T, et al. Adaptive histogram equalization and its variations. Comput Vision Graph Image Process. 1986;39(3):355–68.10.1016/S0734-189X(87)80186-XSearch in Google Scholar

[34] Otsu N. A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern. 1979;9(1):62–6. 10.1109/TSMC.1979.4310076.Search in Google Scholar

[35] Tan M, Chen B, Pang R, Vasudevan V, Le QV. MnasNet: Platform-aware neural architecture search for mobile. 2019 IEEE/CVF Conf Comput Vis Pattern Recognit. 2019 June;2815–23. 10.1109/CVPR.2019.00293.Search in Google Scholar

[36] Yang T, Howard AG, Chen B, Zhang X, Go A, Sze V, et al. NetAdapt: Platform-aware neural network adaptation for mobile applications. Proceedings of the European Conference on Computer Vision (ECCV), 2018, p. 285–300.10.1007/978-3-030-01249-6_18Search in Google Scholar

[37] Hu J, Shen L, Sun G. Squeeze-and-excitation networks. Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit.; Dec. 2018. p. 7132–41. 10.1109/CVPR.2018.00745.Search in Google Scholar

[38] Ross A, Willson VL. One-way anova. In Basic and advanced statistical tests. Rotterdam: SensePublishers; 2017. 10.1007/978-94-6351-086-8_5.Search in Google Scholar

[39] Abdi H, Williams LJ. Turkey’s honestly significant difference (HSD) test. Encyclopedia of research design. Thousand Oaks: Sage; 2010. p. 1–5.Search in Google Scholar

[40] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. 3rd Int. Conf. Learn. Represent. ICLR 2015 – Conf. Track Proc.; Sep. 2014. 10.48550/arxiv.1409.1556.Search in Google Scholar

[41] He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. 2016-December, Dec. 2015. p. 770–8. 10.48550/arxiv.1512.03385.Search in Google Scholar

[42] Huang G, Liu Z, Van Der Maaten L, Weinberger KQ. Densely connected convolutional networks. Proc. - 30th IEEE Conf. Comput. Vis. Pattern Recognition, CVPR 2017. 2017-January, Aug. 2016. p. 2261–9. 10.48550/arxiv.1608.06993.Search in Google Scholar

[43] Szegedy C, Vanhoucke V, Ioffe S, Shlens J, Wojna Z. Rethinking the inception architecture for computer vision. Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit. 2016-December, Dec. 2015. p. 2818–26. 10.48550/arxiv.1512.00567.Search in Google Scholar

[44] Zoph B, Vasudevan V, Shlens J, Le QV. Learning transferable architectures for scalable image recognition. Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit.; Jul. 2017. p. 8697–710. 10.48550/arxiv.1707.07012.Search in Google Scholar
Received: 2023-10-10
Revised: 2024-02-19
Accepted: 2024-02-27
Published Online: 2024-05-15

© 2024 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. Supplementation of P-solubilizing purple nonsulfur bacteria, Rhodopseudomonas palustris improved soil fertility, P nutrient, growth, and yield of Cucumis melo L.
  3. Yield gap variation in rice cultivation in Indonesia
  4. Effects of co-inoculation of indole-3-acetic acid- and ammonia-producing bacteria on plant growth and nutrition, soil elements, and the relationships of soil microbiomes with soil physicochemical parameters
  5. Impact of mulching and planting time on spring-wheat (Triticum aestivum) growth: A combined field experiment and empirical modeling approach
  6. Morphological diversity, correlation studies, and multiple-traits selection for yield and yield components of local cowpea varieties
  7. Participatory on-farm evaluation of new orange-fleshed sweetpotato varieties in Southern Ethiopia
  8. Yield performance and stability analysis of three cultivars of Gayo Arabica coffee across six different environments
  9. Biology of Spodoptera frugiperda (Lepidoptera: Noctuidae) on different types of plants feeds: Potency as a pest on various agricultural plants
  10. Antidiabetic activity of methanolic extract of Hibiscus sabdariffa Linn. fruit in alloxan-induced Swiss albino diabetic mice
  11. Bioinformatics investigation of the effect of volatile and non-volatile compounds of rhizobacteria in inhibiting late embryogenesis abundant protein that induces drought tolerance
  12. Nicotinamide as a biostimulant improves soybean growth and yield
  13. Farmer’s willingness to accept the sustainable zoning-based organic farming development plan: A lesson from Sleman District, Indonesia
  14. Uncovering hidden determinants of millennial farmers’ intentions in running conservation agriculture: An application of the Norm Activation Model
  15. Mediating role of leadership and group capital between human capital component and sustainability of horticultural agribusiness institutions in Indonesia
  16. Biochar technology to increase cassava crop productivity: A study of sustainable agriculture on degraded land
  17. Effect of struvite on the growth of green beans on Mars and Moon regolith simulants
  18. UrbanAgriKG: A knowledge graph on urban agriculture and its embeddings
  19. Provision of loans and credit by cocoa buyers under non-price competition: Cocoa beans market in Ghana
  20. Effectiveness of micro-dosing of lime on selected chemical properties of soil in Banja District, North West, Ethiopia
  21. Effect of weather, nitrogen fertilizer, and biostimulators on the root size and yield components of Hordeum vulgare
  22. Effects of selected biostimulants on qualitative and quantitative parameters of nine cultivars of the genus Capsicum spp.
  23. Growth, yield, and secondary metabolite responses of three shallot cultivars at different watering intervals
  24. Design of drainage channel for effective use of land on fully mechanized sugarcane plantations: A case study at Bone Sugarcane Plantation
  25. Technical feasibility and economic benefit of combined shallot seedlings techniques in Indonesia
  26. Control of Meloidogyne javanica in banana by endophytic bacteria
  27. Comparison of important quality components of red-flesh kiwifruit (Actinidia chinensis) in different locations
  28. Efficiency of rice farming in flood-prone areas of East Java, Indonesia
  29. Comparative analysis of alpine agritourism in Trentino, Tyrol, and South Tyrol: Regional variations and prospects
  30. Detection of Fusarium spp. infection in potato (Solanum tuberosum L.) during postharvest storage through visible–near-infrared and shortwave–near-infrared reflectance spectroscopy
  31. Forage yield, seed, and forage qualitative traits evaluation by determining the optimal forage harvesting stage in dual-purpose cultivation in safflower varieties (Carthamus tinctorius L.)
  32. The influence of tourism on the development of urban space: Comparison in Hanoi, Danang, and Ho Chi Minh City
  33. Optimum intra-row spacing and clove size for the economical production of garlic (Allium sativum L.) in Northwestern Highlands of Ethiopia
  34. The role of organic rice farm income on farmer household welfare: Evidence from Yogyakarta, Indonesia
  35. Exploring innovative food in a developing country: Edible insects as a sustainable option
  36. Genotype by environment interaction and performance stability of common bean (Phaseolus vulgaris L.) cultivars grown in Dawuro zone, Southwestern Ethiopia
  37. Factors influencing green, environmentally-friendly consumer behaviour
  38. Factors affecting coffee farmers’ access to financial institutions: The case of Bandung Regency, Indonesia
  39. Morphological and yield trait-based evaluation and selection of chili (Capsicum annuum L.) genotypes suitable for both summer and winter seasons
  40. Sustainability analysis and decision-making strategy for swamp buffalo (Bubalus bubalis carabauesis) conservation in Jambi Province, Indonesia
  41. Understanding factors affecting rice purchasing decisions in Indonesia: Does rice brand matter?
  42. An implementation of an extended theory of planned behavior to investigate consumer behavior on hygiene sanitation-certified livestock food products
  43. Information technology adoption in Indonesia’s small-scale dairy farms
  44. Draft genome of a biological control agent against Bipolaris sorokiniana, the causal phytopathogen of spot blotch in wheat (Triticum turgidum L. subsp. durum): Bacillus inaquosorum TSO22
  45. Assessment of the recurrent mutagenesis efficacy of sesame crosses followed by isolation and evaluation of promising genetic resources for use in future breeding programs
  46. Fostering cocoa industry resilience: A collaborative approach to managing farm gate price fluctuations in West Sulawesi, Indonesia
  47. Field investigation of component failures for selected farm machinery used in small rice farming operations
  48. Near-infrared technology in agriculture: Rapid, simultaneous, and non-destructive determination of inner quality parameters on intact coffee beans
  49. The synergistic application of sucrose and various LED light exposures to enhance the in vitro growth of Stevia rebaudiana (Bertoni)
  50. Weather index-based agricultural insurance for flower farmers: Willingness to pay, sales, and profitability perspectives
  51. Meta-analysis of dietary Bacillus spp. on serum biochemical and antioxidant status and egg quality of laying hens
  52. Biochemical characterization of trypsin from Indonesian skipjack tuna (Katsuwonus pelamis) viscera
  53. Determination of C-factor for conventional cultivation and soil conservation technique used in hop gardens
  54. Empowering farmers: Unveiling the economic impacts of contract farming on red chilli farmers’ income in Magelang District, Indonesia
  55. Evaluating salt tolerance in fodder crops: A field experiment in the dry land
  56. Labor productivity of lowland rice (Oryza sativa L.) farmers in Central Java Province, Indonesia
  57. Cropping systems and production assessment in southern Myanmar: Informing strategic interventions
  58. The effect of biostimulants and red mud on the growth and yield of shallots in post-unlicensed gold mining soil
  59. Effects of dietary Adansonia digitata L. (baobab) seed meal on growth performance and carcass characteristics of broiler chickens: A systematic review and meta-analysis
  60. Analysis and structural characterization of the vid-pisco market
  61. Pseudomonas fluorescens SP007s enhances defense responses against the soybean bacterial pustule caused by Xanthomonas axonopodis pv. glycines
  62. A brief investigation on the prospective of co-composted biochar as a fertilizer for Zucchini plants cultivated in arid sandy soil
  63. Supply chain efficiency of red chilies in the production center of Sleman Indonesia based on performance measurement system
  64. Investment development path for developed economies: Is agriculture different?
  65. Power relations among actors in laying hen business in Indonesia: A MACTOR analysis
  66. High-throughput digital imaging and detection of morpho-physiological traits in tomato plants under drought
  67. Converting compression ignition engine to dual-fuel (diesel + CNG) engine and experimentally investigating its performance and emissions
  68. Structuration, risk management, and institutional dynamics in resolving palm oil conflicts
  69. Spacing strategies for enhancing drought resilience and yield in maize agriculture
  70. Composition and quality of winter annual agrestal and ruderal herbages of two different land-use types
  71. Investigating Spodoptera spp. diversity, percentage of attack, and control strategies in the West Java, Indonesia, corn cultivation
  72. Yield stability of biofertilizer treatments to soybean in the rainy season based on the GGE biplot
  73. Evaluating agricultural yield and economic implications of varied irrigation depths on maize yield in semi-arid environments, at Birfarm, Upper Blue Nile, Ethiopia
  74. Chemometrics for mapping the spatial nitrate distribution on the leaf lamina of fenugreek grown under varying nitrogenous fertilizer doses
  75. Pomegranate peel ethanolic extract: A promising natural antioxidant, antimicrobial agent, and novel approach to mitigate rancidity in used edible oils
  76. Transformative learning and engagement with organic farming: Lessons learned from Indonesia
  77. Tourism in rural areas as a broader concept: Some insights from the Portuguese reality
  78. Assessment enhancing drought tolerance in henna (Lawsonia inermis L.) ecotypes through sodium nitroprusside foliar application
  79. Edible insects: A survey about perceptions regarding possible beneficial health effects and safety concerns among adult citizens from Portugal and Romania
  80. Phenological stages analysis in peach trees using electronic nose
  81. Harvest date and salicylic acid impact on peanut (Arachis hypogaea L.) properties under different humidity conditions
  82. Hibiscus sabdariffa L. petal biomass: A green source of nanoparticles of multifarious potential
  83. Use of different vegetation indices for the evaluation of the kinetics of the cherry tomato (Solanum lycopersicum var. cerasiforme) growth based on multispectral images by UAV
  84. First evidence of microplastic pollution in mangrove sediments and its ingestion by coral reef fish: Case study in Biawak Island, Indonesia
  85. Physical and textural properties and sensory acceptability of wheat bread partially incorporated with unripe non-commercial banana cultivars
  86. Cereibacter sphaeroides ST16 and ST26 were used to solubilize insoluble P forms to improve P uptake, growth, and yield of rice in acidic and extreme saline soil
  87. Avocado peel by-product in cattle diets and supplementation with oregano oil and effects on production, carcass, and meat quality
  88. Optimizing inorganic blended fertilizer application for the maximum grain yield and profitability of bread wheat and food barley in Dawuro Zone, Southwest Ethiopia
  89. The acceptance of social media as a channel of communication and livestock information for sheep farmers
  90. Adaptation of rice farmers to aging in Thailand
  91. Combined use of improved maize hybrids and nitrogen application increases grain yield of maize, under natural Striga hermonthica infestation
  92. From aquatic to terrestrial: An examination of plant diversity and ecological shifts
  93. Statistical modelling of a tractor tractive performance during ploughing operation on a tropical Alfisol
  94. Participation in artisanal diamond mining and food security: A case study of Kasai Oriental in DR Congo
  95. Assessment and multi-scenario simulation of ecosystem service values in Southwest China’s mountainous and hilly region
  96. Analysis of agricultural emissions and economic growth in Europe in search of ecological balance
  97. Bacillus thuringiensis strains with high insecticidal activity against insect larvae of the orders Coleoptera and Lepidoptera
  98. Technical efficiency of sugarcane farming in East Java, Indonesia: A bootstrap data envelopment analysis
  99. Comparison between mycobiota diversity and fungi and mycotoxin contamination of maize and wheat
  100. Evaluation of cultivation technology package and corn variety based on agronomy characters and leaf green indices
  101. Exploring the association between the consumption of beverages, fast foods, sweets, fats, and oils and the risk of gastric and pancreatic cancers: Findings from case–control study
  102. Phytochemical composition and insecticidal activity of Acokanthera oblongifolia (Hochst.) Benth & Hook.f. ex B.D.Jacks. extract on life span and biological aspects of Spodoptera littoralis (Biosd.)
  103. Land use management solutions in response to climate change: Case study in the central coastal areas of Vietnam
  104. Evaluation of coffee pulp as a feed ingredient for ruminants: A meta-analysis
  105. Interannual variations of normalized difference vegetation index and potential evapotranspiration and their relationship in the Baghdad area
  106. Harnessing synthetic microbial communities with nitrogen-fixing activity to promote rice growth
  107. Agronomic and economic benefits of rice–sweetpotato rotation in lowland rice cropping systems in Uganda
  108. Response of potato tuber as an effect of the N-fertilizer and paclobutrazol application in medium altitude
  109. Bridging the gap: The role of geographic proximity in enhancing seed sustainability in Bandung District
  110. Evaluation of Abrams curve in agricultural sector using the NARDL approach
  111. Challenges and opportunities for young farmers in the implementation of the Rural Development Program 2014–2020 of the Republic of Croatia
  112. Yield stability of ten common bean (Phaseolus vulgaris L.) genotypes at different sowing dates in Lubumbashi, South-East of DR Congo
  113. Effects of encapsulation and combining probiotics with different nitrate forms on methane emission and in vitro rumen fermentation characteristics
  114. Phytochemical analysis of Bienertia sinuspersici extract and its antioxidant and antimicrobial activities
  115. Evaluation of relative drought tolerance of grapevines by leaf fluorescence parameters
  116. Yield assessment of new streak-resistant topcross maize hybrids in Benin
  117. Improvement of cocoa powder properties through ultrasonic- and microwave-assisted alkalization
  118. Potential of ecoenzymes made from nutmeg (Myristica fragrans) leaf and pulp waste as bioinsecticides for Periplaneta americana
  119. Analysis of farm performance to realize the sustainability of organic cabbage vegetable farming in Getasan Semarang, Indonesia
  120. Revealing the influences of organic amendment-derived dissolved organic matter on growth and nutrient accumulation in lettuce seedlings (Lactuca sativa L.)
  121. Identification of viruses infecting sweetpotato (Ipomoea batatas Lam.) in Benin
  122. Assessing the soil physical and chemical properties of long-term pomelo orchard based on tree growth
  123. Investigating access and use of digital tools for agriculture among rural farmers: A case study of Nkomazi Municipality, South Africa
  124. Does sex influence the impact of dietary vitD3 and UVB light on performance parameters and welfare indicators of broilers?
  125. Design of intelligent sprayer control for an autonomous farming drone using a multiclass support vector machine
  126. Deciphering salt-responsive NB-ARC genes in rice transcriptomic data: A bioinformatics approach with gene expression validation
  127. Review Articles
  128. Impact of nematode infestation in livestock production and the role of natural feed additives – A review
  129. Role of dietary fats in reproductive, health, and nutritional benefits in farm animals: A review
  130. Climate change and adaptive strategies on viticulture (Vitis spp.)
  131. The false tiger of almond, Monosteira unicostata (Hemiptera: Tingidae): Biology, ecology, and control methods
  132. A systematic review on potential analogy of phytobiomass and soil carbon evaluation methods: Ethiopia insights
  133. A review of storage temperature and relative humidity effects on shelf life and quality of mango (Mangifera indica L.) fruit and implications for nutrition insecurity in Ethiopia
  134. Green extraction of nutmeg (Myristica fragrans) phytochemicals: Prospective strategies and roadblocks
  135. Potential influence of nitrogen fertilizer rates on yield and yield components of carrot (Dacus carota L.) in Ethiopia: Systematic review
  136. Corn silk: A promising source of antimicrobial compounds for health and wellness
  137. State and contours of research on roselle (Hibiscus sabdariffa L.) in Africa
  138. The potential of phosphorus-solubilizing purple nonsulfur bacteria in agriculture: Present and future perspectives
  139. Minor millets: Processing techniques and their nutritional and health benefits
  140. Meta-analysis of reproductive performance of improved dairy cattle under Ethiopian environmental conditions
  141. Review on enhancing the efficiency of fertilizer utilization: Strategies for optimal nutrient management
  142. The nutritional, phytochemical composition, and utilisation of different parts of maize: A comparative analysis
  143. Motivations for farmers’ participation in agri-environmental scheme in the EU, literature review
  144. Evolution of climate-smart agriculture research: A science mapping exploration and network analysis
  145. Short Communications
  146. Music enrichment improves the behavior and leukocyte profile of dairy cattle
  147. Effect of pruning height and organic fertilization on the morphological and productive characteristics of Moringa oleifera Lam. in the Peruvian dry tropics
  148. Corrigendum
  149. Corrigendum to “Bioinformatics investigation of the effect of volatile and non-volatile compounds of rhizobacteria in inhibiting late embryogenesis abundant protein that induces drought tolerance”
  150. Corrigendum to “Composition and quality of winter annual agrestal and ruderal herbages of two different land-use types”
  151. Special issue: Smart Agriculture System for Sustainable Development: Methods and Practices
  152. Construction of a sustainable model to predict the moisture content of porang powder (Amorphophallus oncophyllus) based on pointed-scan visible near-infrared spectroscopy
  153. FruitVision: A deep learning based automatic fruit grading system
  154. Energy harvesting and ANFIS modeling of a PVDF/GO-ZNO piezoelectric nanogenerator on a UAV
  155. Effects of stress hormones on digestibility and performance in cattle: A review
  156. Special Issue of The 4th International Conference on Food Science and Engineering (ICFSE) 2022 - Part II
  157. Assessment of omega-3 and omega-6 fatty acid profiles and ratio of omega-6/omega-3 of white eggs produced by laying hens fed diets enriched with omega-3 rich vegetable oil
  158. Special Issue on FCEM - International Web Conference on Food Choice & Eating Motivation - Part II
  159. Special Issue on FCEM – International Web Conference on Food Choice & Eating Motivation: Message from the editor
  160. Fruit and vegetable consumption: Study involving Portuguese and French consumers
  161. Knowledge about consumption of milk: Study involving consumers from two European Countries – France and Portugal
https://doi.org/10.1515/opag-2022-0276
Keywords for this article
machine learning; computer vision; deep learning; fruit grading; agriculture
Creative Commons
BY 4.0

Articles in the same Issue

  1. Regular Articles
  2. Supplementation of P-solubilizing purple nonsulfur bacteria, Rhodopseudomonas palustris improved soil fertility, P nutrient, growth, and yield of Cucumis melo L.
  3. Yield gap variation in rice cultivation in Indonesia
  4. Effects of co-inoculation of indole-3-acetic acid- and ammonia-producing bacteria on plant growth and nutrition, soil elements, and the relationships of soil microbiomes with soil physicochemical parameters
  5. Impact of mulching and planting time on spring-wheat (Triticum aestivum) growth: A combined field experiment and empirical modeling approach
  6. Morphological diversity, correlation studies, and multiple-traits selection for yield and yield components of local cowpea varieties
  7. Participatory on-farm evaluation of new orange-fleshed sweetpotato varieties in Southern Ethiopia
  8. Yield performance and stability analysis of three cultivars of Gayo Arabica coffee across six different environments
  9. Biology of Spodoptera frugiperda (Lepidoptera: Noctuidae) on different types of plants feeds: Potency as a pest on various agricultural plants
  10. Antidiabetic activity of methanolic extract of Hibiscus sabdariffa Linn. fruit in alloxan-induced Swiss albino diabetic mice
  11. Bioinformatics investigation of the effect of volatile and non-volatile compounds of rhizobacteria in inhibiting late embryogenesis abundant protein that induces drought tolerance
  12. Nicotinamide as a biostimulant improves soybean growth and yield
  13. Farmer’s willingness to accept the sustainable zoning-based organic farming development plan: A lesson from Sleman District, Indonesia
  14. Uncovering hidden determinants of millennial farmers’ intentions in running conservation agriculture: An application of the Norm Activation Model
  15. Mediating role of leadership and group capital between human capital component and sustainability of horticultural agribusiness institutions in Indonesia
  16. Biochar technology to increase cassava crop productivity: A study of sustainable agriculture on degraded land
  17. Effect of struvite on the growth of green beans on Mars and Moon regolith simulants
  18. UrbanAgriKG: A knowledge graph on urban agriculture and its embeddings
  19. Provision of loans and credit by cocoa buyers under non-price competition: Cocoa beans market in Ghana
  20. Effectiveness of micro-dosing of lime on selected chemical properties of soil in Banja District, North West, Ethiopia
  21. Effect of weather, nitrogen fertilizer, and biostimulators on the root size and yield components of Hordeum vulgare
  22. Effects of selected biostimulants on qualitative and quantitative parameters of nine cultivars of the genus Capsicum spp.
  23. Growth, yield, and secondary metabolite responses of three shallot cultivars at different watering intervals
  24. Design of drainage channel for effective use of land on fully mechanized sugarcane plantations: A case study at Bone Sugarcane Plantation
  25. Technical feasibility and economic benefit of combined shallot seedlings techniques in Indonesia
  26. Control of Meloidogyne javanica in banana by endophytic bacteria
  27. Comparison of important quality components of red-flesh kiwifruit (Actinidia chinensis) in different locations
  28. Efficiency of rice farming in flood-prone areas of East Java, Indonesia
  29. Comparative analysis of alpine agritourism in Trentino, Tyrol, and South Tyrol: Regional variations and prospects
  30. Detection of Fusarium spp. infection in potato (Solanum tuberosum L.) during postharvest storage through visible–near-infrared and shortwave–near-infrared reflectance spectroscopy
  31. Forage yield, seed, and forage qualitative traits evaluation by determining the optimal forage harvesting stage in dual-purpose cultivation in safflower varieties (Carthamus tinctorius L.)
  32. The influence of tourism on the development of urban space: Comparison in Hanoi, Danang, and Ho Chi Minh City
  33. Optimum intra-row spacing and clove size for the economical production of garlic (Allium sativum L.) in Northwestern Highlands of Ethiopia
  34. The role of organic rice farm income on farmer household welfare: Evidence from Yogyakarta, Indonesia
  35. Exploring innovative food in a developing country: Edible insects as a sustainable option
  36. Genotype by environment interaction and performance stability of common bean (Phaseolus vulgaris L.) cultivars grown in Dawuro zone, Southwestern Ethiopia
  37. Factors influencing green, environmentally-friendly consumer behaviour
  38. Factors affecting coffee farmers’ access to financial institutions: The case of Bandung Regency, Indonesia
  39. Morphological and yield trait-based evaluation and selection of chili (Capsicum annuum L.) genotypes suitable for both summer and winter seasons
  40. Sustainability analysis and decision-making strategy for swamp buffalo (Bubalus bubalis carabauesis) conservation in Jambi Province, Indonesia
  41. Understanding factors affecting rice purchasing decisions in Indonesia: Does rice brand matter?
  42. An implementation of an extended theory of planned behavior to investigate consumer behavior on hygiene sanitation-certified livestock food products
  43. Information technology adoption in Indonesia’s small-scale dairy farms
  44. Draft genome of a biological control agent against Bipolaris sorokiniana, the causal phytopathogen of spot blotch in wheat (Triticum turgidum L. subsp. durum): Bacillus inaquosorum TSO22
  45. Assessment of the recurrent mutagenesis efficacy of sesame crosses followed by isolation and evaluation of promising genetic resources for use in future breeding programs
  46. Fostering cocoa industry resilience: A collaborative approach to managing farm gate price fluctuations in West Sulawesi, Indonesia
  47. Field investigation of component failures for selected farm machinery used in small rice farming operations
  48. Near-infrared technology in agriculture: Rapid, simultaneous, and non-destructive determination of inner quality parameters on intact coffee beans
  49. The synergistic application of sucrose and various LED light exposures to enhance the in vitro growth of Stevia rebaudiana (Bertoni)
  50. Weather index-based agricultural insurance for flower farmers: Willingness to pay, sales, and profitability perspectives
  51. Meta-analysis of dietary Bacillus spp. on serum biochemical and antioxidant status and egg quality of laying hens
  52. Biochemical characterization of trypsin from Indonesian skipjack tuna (Katsuwonus pelamis) viscera
  53. Determination of C-factor for conventional cultivation and soil conservation technique used in hop gardens
  54. Empowering farmers: Unveiling the economic impacts of contract farming on red chilli farmers’ income in Magelang District, Indonesia
  55. Evaluating salt tolerance in fodder crops: A field experiment in the dry land
  56. Labor productivity of lowland rice (Oryza sativa L.) farmers in Central Java Province, Indonesia
  57. Cropping systems and production assessment in southern Myanmar: Informing strategic interventions
  58. The effect of biostimulants and red mud on the growth and yield of shallots in post-unlicensed gold mining soil
  59. Effects of dietary Adansonia digitata L. (baobab) seed meal on growth performance and carcass characteristics of broiler chickens: A systematic review and meta-analysis
  60. Analysis and structural characterization of the vid-pisco market
  61. Pseudomonas fluorescens SP007s enhances defense responses against the soybean bacterial pustule caused by Xanthomonas axonopodis pv. glycines
  62. A brief investigation on the prospective of co-composted biochar as a fertilizer for Zucchini plants cultivated in arid sandy soil
  63. Supply chain efficiency of red chilies in the production center of Sleman Indonesia based on performance measurement system
  64. Investment development path for developed economies: Is agriculture different?
  65. Power relations among actors in laying hen business in Indonesia: A MACTOR analysis
  66. High-throughput digital imaging and detection of morpho-physiological traits in tomato plants under drought
  67. Converting compression ignition engine to dual-fuel (diesel + CNG) engine and experimentally investigating its performance and emissions
  68. Structuration, risk management, and institutional dynamics in resolving palm oil conflicts
  69. Spacing strategies for enhancing drought resilience and yield in maize agriculture
  70. Composition and quality of winter annual agrestal and ruderal herbages of two different land-use types
  71. Investigating Spodoptera spp. diversity, percentage of attack, and control strategies in the West Java, Indonesia, corn cultivation
  72. Yield stability of biofertilizer treatments to soybean in the rainy season based on the GGE biplot
  73. Evaluating agricultural yield and economic implications of varied irrigation depths on maize yield in semi-arid environments, at Birfarm, Upper Blue Nile, Ethiopia
  74. Chemometrics for mapping the spatial nitrate distribution on the leaf lamina of fenugreek grown under varying nitrogenous fertilizer doses
  75. Pomegranate peel ethanolic extract: A promising natural antioxidant, antimicrobial agent, and novel approach to mitigate rancidity in used edible oils
  76. Transformative learning and engagement with organic farming: Lessons learned from Indonesia
  77. Tourism in rural areas as a broader concept: Some insights from the Portuguese reality
  78. Assessment enhancing drought tolerance in henna (Lawsonia inermis L.) ecotypes through sodium nitroprusside foliar application
  79. Edible insects: A survey about perceptions regarding possible beneficial health effects and safety concerns among adult citizens from Portugal and Romania
  80. Phenological stages analysis in peach trees using electronic nose
  81. Harvest date and salicylic acid impact on peanut (Arachis hypogaea L.) properties under different humidity conditions
  82. Hibiscus sabdariffa L. petal biomass: A green source of nanoparticles of multifarious potential
  83. Use of different vegetation indices for the evaluation of the kinetics of the cherry tomato (Solanum lycopersicum var. cerasiforme) growth based on multispectral images by UAV
  84. First evidence of microplastic pollution in mangrove sediments and its ingestion by coral reef fish: Case study in Biawak Island, Indonesia
  85. Physical and textural properties and sensory acceptability of wheat bread partially incorporated with unripe non-commercial banana cultivars
  86. Cereibacter sphaeroides ST16 and ST26 were used to solubilize insoluble P forms to improve P uptake, growth, and yield of rice in acidic and extreme saline soil
  87. Avocado peel by-product in cattle diets and supplementation with oregano oil and effects on production, carcass, and meat quality
  88. Optimizing inorganic blended fertilizer application for the maximum grain yield and profitability of bread wheat and food barley in Dawuro Zone, Southwest Ethiopia
  89. The acceptance of social media as a channel of communication and livestock information for sheep farmers
  90. Adaptation of rice farmers to aging in Thailand
  91. Combined use of improved maize hybrids and nitrogen application increases grain yield of maize, under natural Striga hermonthica infestation
  92. From aquatic to terrestrial: An examination of plant diversity and ecological shifts
  93. Statistical modelling of a tractor tractive performance during ploughing operation on a tropical Alfisol
  94. Participation in artisanal diamond mining and food security: A case study of Kasai Oriental in DR Congo
  95. Assessment and multi-scenario simulation of ecosystem service values in Southwest China’s mountainous and hilly region
  96. Analysis of agricultural emissions and economic growth in Europe in search of ecological balance
  97. Bacillus thuringiensis strains with high insecticidal activity against insect larvae of the orders Coleoptera and Lepidoptera
  98. Technical efficiency of sugarcane farming in East Java, Indonesia: A bootstrap data envelopment analysis
  99. Comparison between mycobiota diversity and fungi and mycotoxin contamination of maize and wheat
  100. Evaluation of cultivation technology package and corn variety based on agronomy characters and leaf green indices
  101. Exploring the association between the consumption of beverages, fast foods, sweets, fats, and oils and the risk of gastric and pancreatic cancers: Findings from case–control study
  102. Phytochemical composition and insecticidal activity of Acokanthera oblongifolia (Hochst.) Benth & Hook.f. ex B.D.Jacks. extract on life span and biological aspects of Spodoptera littoralis (Biosd.)
  103. Land use management solutions in response to climate change: Case study in the central coastal areas of Vietnam
  104. Evaluation of coffee pulp as a feed ingredient for ruminants: A meta-analysis
  105. Interannual variations of normalized difference vegetation index and potential evapotranspiration and their relationship in the Baghdad area
  106. Harnessing synthetic microbial communities with nitrogen-fixing activity to promote rice growth
  107. Agronomic and economic benefits of rice–sweetpotato rotation in lowland rice cropping systems in Uganda
  108. Response of potato tuber as an effect of the N-fertilizer and paclobutrazol application in medium altitude
  109. Bridging the gap: The role of geographic proximity in enhancing seed sustainability in Bandung District
  110. Evaluation of Abrams curve in agricultural sector using the NARDL approach
  111. Challenges and opportunities for young farmers in the implementation of the Rural Development Program 2014–2020 of the Republic of Croatia
  112. Yield stability of ten common bean (Phaseolus vulgaris L.) genotypes at different sowing dates in Lubumbashi, South-East of DR Congo
  113. Effects of encapsulation and combining probiotics with different nitrate forms on methane emission and in vitro rumen fermentation characteristics
  114. Phytochemical analysis of Bienertia sinuspersici extract and its antioxidant and antimicrobial activities
  115. Evaluation of relative drought tolerance of grapevines by leaf fluorescence parameters
  116. Yield assessment of new streak-resistant topcross maize hybrids in Benin
  117. Improvement of cocoa powder properties through ultrasonic- and microwave-assisted alkalization
  118. Potential of ecoenzymes made from nutmeg (Myristica fragrans) leaf and pulp waste as bioinsecticides for Periplaneta americana
  119. Analysis of farm performance to realize the sustainability of organic cabbage vegetable farming in Getasan Semarang, Indonesia
  120. Revealing the influences of organic amendment-derived dissolved organic matter on growth and nutrient accumulation in lettuce seedlings (Lactuca sativa L.)
  121. Identification of viruses infecting sweetpotato (Ipomoea batatas Lam.) in Benin
  122. Assessing the soil physical and chemical properties of long-term pomelo orchard based on tree growth
  123. Investigating access and use of digital tools for agriculture among rural farmers: A case study of Nkomazi Municipality, South Africa
  124. Does sex influence the impact of dietary vitD3 and UVB light on performance parameters and welfare indicators of broilers?
  125. Design of intelligent sprayer control for an autonomous farming drone using a multiclass support vector machine
  126. Deciphering salt-responsive NB-ARC genes in rice transcriptomic data: A bioinformatics approach with gene expression validation
  127. Review Articles
  128. Impact of nematode infestation in livestock production and the role of natural feed additives – A review
  129. Role of dietary fats in reproductive, health, and nutritional benefits in farm animals: A review
  130. Climate change and adaptive strategies on viticulture (Vitis spp.)
  131. The false tiger of almond, Monosteira unicostata (Hemiptera: Tingidae): Biology, ecology, and control methods
  132. A systematic review on potential analogy of phytobiomass and soil carbon evaluation methods: Ethiopia insights
  133. A review of storage temperature and relative humidity effects on shelf life and quality of mango (Mangifera indica L.) fruit and implications for nutrition insecurity in Ethiopia
  134. Green extraction of nutmeg (Myristica fragrans) phytochemicals: Prospective strategies and roadblocks
  135. Potential influence of nitrogen fertilizer rates on yield and yield components of carrot (Dacus carota L.) in Ethiopia: Systematic review
  136. Corn silk: A promising source of antimicrobial compounds for health and wellness
  137. State and contours of research on roselle (Hibiscus sabdariffa L.) in Africa
  138. The potential of phosphorus-solubilizing purple nonsulfur bacteria in agriculture: Present and future perspectives
  139. Minor millets: Processing techniques and their nutritional and health benefits
  140. Meta-analysis of reproductive performance of improved dairy cattle under Ethiopian environmental conditions
  141. Review on enhancing the efficiency of fertilizer utilization: Strategies for optimal nutrient management
  142. The nutritional, phytochemical composition, and utilisation of different parts of maize: A comparative analysis
  143. Motivations for farmers’ participation in agri-environmental scheme in the EU, literature review
  144. Evolution of climate-smart agriculture research: A science mapping exploration and network analysis
  145. Short Communications
  146. Music enrichment improves the behavior and leukocyte profile of dairy cattle
  147. Effect of pruning height and organic fertilization on the morphological and productive characteristics of Moringa oleifera Lam. in the Peruvian dry tropics
  148. Corrigendum
  149. Corrigendum to “Bioinformatics investigation of the effect of volatile and non-volatile compounds of rhizobacteria in inhibiting late embryogenesis abundant protein that induces drought tolerance”
  150. Corrigendum to “Composition and quality of winter annual agrestal and ruderal herbages of two different land-use types”
  151. Special issue: Smart Agriculture System for Sustainable Development: Methods and Practices
  152. Construction of a sustainable model to predict the moisture content of porang powder (Amorphophallus oncophyllus) based on pointed-scan visible near-infrared spectroscopy
  153. FruitVision: A deep learning based automatic fruit grading system
  154. Energy harvesting and ANFIS modeling of a PVDF/GO-ZNO piezoelectric nanogenerator on a UAV
  155. Effects of stress hormones on digestibility and performance in cattle: A review
  156. Special Issue of The 4th International Conference on Food Science and Engineering (ICFSE) 2022 - Part II
  157. Assessment of omega-3 and omega-6 fatty acid profiles and ratio of omega-6/omega-3 of white eggs produced by laying hens fed diets enriched with omega-3 rich vegetable oil
  158. Special Issue on FCEM - International Web Conference on Food Choice & Eating Motivation - Part II
  159. Special Issue on FCEM – International Web Conference on Food Choice & Eating Motivation: Message from the editor
  160. Fruit and vegetable consumption: Study involving Portuguese and French consumers
  161. Knowledge about consumption of milk: Study involving consumers from two European Countries – France and Portugal
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 5.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/opag-2022-0276/html
Scroll to top button