Fruit Image Classification Model Based on MobileNetV2 with Deep Transfer Learning Technique

Sustainability, Год журнала: 2023, Номер 15(3), С. 1906 - 1906

Опубликована: Янв. 19, 2023

Due to the rapid emergence and evolution of AI applications, utilization smart imaging devices has increased significantly. Researchers have started using deep learning models, such as CNN, for image classification. Unlike traditional which require a lot features perform well, CNN does not any handcrafted well. It uses numerous filters, extract required from images automatically One issues in horticulture industry is fruit classification, requires an expert with experience. To overcome this issue automated system can classify different types fruits without need human effort. In study, dataset total 26,149 40 was used experimentation. The training test set were randomly recreated divided into ratio 3:1. experiment introduces customized head five layers MobileNetV2 architecture. classification layer model replaced by head, produced modified version called TL-MobileNetV2. addition, transfer retain pre-trained model. TL-MobileNetV2 achieves accuracy 99%, 3% higher than MobileNetV2, equal error rate just 1%. Compared AlexNet, VGG16, InceptionV3, ResNet, better 8, 11, 6, 10%, respectively. Furthermore, obtained 99% precision, recall, F1-score. be concluded that plays big part achieving results, dropout technique helps reduce overfitting learning.

Язык: Английский

---

Enhancing Image Annotation Technique of Fruit Classification Using a Deep Learning Approach

Sustainability, Год журнала: 2023, Номер 15(2), С. 901 - 901

Опубликована: Янв. 4, 2023

An accurate image retrieval technique is required due to the rapidly increasing number of images. It important implement annotation techniques that are fast, simple, and, most importantly, automatically annotate. Image has recently received much attention massive rise in data volume. Focusing on agriculture field, this study implements automatic annotation, namely, a repetitive task technique, classify ripeness oil palm fruit and recognize variety fruits. This approach assists farmers enhance classification methods increase their production. proposes simple effective models using deep learning with You Only Look Once (YOLO) versions. The were developed through transfer where dataset was trained 100 images 400 RGB Model performance accuracy annotating 3500 fruits examined. results show successfully annotated large accurately. mAP result achieved for 98.7% 99.5%.

Язык: Английский

---

Artificial Intelligence in Agriculture: Benefits, Challenges, and Trends

Applied Sciences, Год журнала: 2023, Номер 13(13), С. 7405 - 7405

Опубликована: Июнь 22, 2023

The world’s population has reached 8 billion and is projected to reach 9.7 by 2050, increasing the demand for food production. Artificial intelligence (AI) technologies that optimize resources increase productivity are vital in an environment tensions supply chain increasingly frequent weather events. This study performed a systemic review of literature using Preferred Reporting Items Systematic Reviews Meta-Analyses (PRISMA) methodology on artificial applied agriculture. It retrieved 906 relevant studies from five electronic databases selected 176 bibliometric analysis. quality appraisal step 17 analysis benefits, challenges, trends AI used work showed evolution area with increased publications over last years, more than 20 different techniques analyzed, machine learning, convolutional neural networks, IoT, big data, robotics, computer vision being most technologies. Considering worldwide scope, countries highlighted were India, China, USA. Agricultural sectors included crop management prediction disease pest management. Finally, it presented challenges promising when considering future directions

Язык: Английский

---

Image Acquisition, Preprocessing and Classification of Citrus Fruit Diseases: A Systematic Literature Review

Sustainability, Год журнала: 2023, Номер 15(12), С. 9643 - 9643

Опубликована: Июнь 15, 2023

Different kinds of techniques are evaluated and analyzed for various classification models the detection diseases citrus fruits. This paper aims to systematically review papers that focus on prediction, detection, fruit have employed machine learning, deep statistical techniques. Additionally, this explores present state art concept image acquisition, digital processing, feature extraction, approaches, each one is discussed separately. A total 78 selected after applying primary selection criteria, inclusion/exclusion quality assessment criteria. We observe following widely used in studies: hyperspectral imaging systems acquisition process, thresholding support vector (SVM) as learning (ML) models, convolutional neural network (CNN) architectures principal component analysis (PCA) a model, accuracy evaluation parameters. Moreover, color most popularly RGB space. From studies performed comparative analyses, we find best outperformed other their respective categories follows: SVM among ML methods, ANN networks, CNN linear discriminant (LDA) techniques.This study concludes with meta-analysis, limitations, future research directions.

Язык: Английский

---

An Artificial Intelligence-Based Stacked Ensemble Approach for Prediction of Protein Subcellular Localization in Confocal Microscopy Images

Sustainability, Год журнала: 2023, Номер 15(2), С. 1695 - 1695

Опубликована: Янв. 16, 2023

Predicting subcellular protein localization has become a popular topic due to its utility in understanding disease mechanisms and developing innovative drugs. With the rapid advancement of automated microscopic imaging technology, approaches using bio-images for have gained lot interest. The Human Protein Atlas (HPA) project is macro-initiative that aims map human proteome utilizing antibody-based proteomics related c. Millions images been tagged with single or multiple labels HPA database. However, fewer techniques predicting location proteins devised, majority them relying on automatic single-label classification. As result, there need an sustainable system capable multi-label classification Deep learning presents potential option labeling protein’s localization, given vast image number generated by high-content microscopy fact manual both time-consuming error-prone. Hence, this research use ensemble technique improvement performance existing state-of-art convolutional neural networks pretrained models were applied; finally, stacked ensemble-based deep model was presented, which delivers more reliable robust classifier. F1-score, precision, recall used evaluation proposed model’s efficiency. In addition, comparison conducted respect method. results show strategy performed exponentially well images, recall, F1-score 0.70, 0.72, 0.71, respectively.

Язык: Английский

---

Harnessing the Power of Transfer Learning in Sunflower Disease Detection: A Comparative Study

Agriculture, Год журнала: 2023, Номер 13(8), С. 1479 - 1479

Опубликована: Июль 26, 2023

Sunflower is an important crop that susceptible to various diseases, which can significantly impact yield and quality. Early accurate detection of these diseases crucial for implementing appropriate management strategies. In recent years, deep learning techniques have shown promising results in the field disease classification using image data. This study presents a comparative analysis different deep-learning models sunflower diseases. five widely used models, namely AlexNet, VGG16, InceptionV3, MobileNetV3, EfficientNet were trained evaluated dataset images. The performance each model was measured terms precision, recall, F1-score, accuracy. experimental demonstrated all achieved high accuracy values classification. Among EfficientNetB3 exhibited highest 0.979. whereas other ALexNet, InceptionV3 MobileNetV3 0.865, 0.965, 0.954 0.969 respectively. Based on analysis, it be concluded are effective highlight potential early classification, assist farmers agronomists timely Furthermore, findings suggest like could preferred choices due their relatively fewer training epochs.

Язык: Английский

---

Enhanced corn seed disease classification: leveraging MobileNetV2 with feature augmentation and transfer learning

Frontiers in Applied Mathematics and Statistics, Год журнала: 2024, Номер 9

Опубликована: Янв. 3, 2024

In the era of advancing artificial intelligence (AI), its application in agriculture has become increasingly pivotal. This study explores integration AI for discriminative classification corn diseases, addressing need efficient agricultural practices. Leveraging a comprehensive dataset, encompasses 21,662 images categorized into four classes: Broken, Discolored, Silk cut, and Pure. The proposed model, an enhanced iteration MobileNetV2, strategically incorporates additional layers—Average Pooling, Flatten, Dense, Dropout, softmax—augmenting feature extraction capabilities. Model tuning techniques, including data augmentation, adaptive learning rate, model checkpointing, dropout, transfer learning, fortify model's efficiency. Results showcase exceptional performance, achieving accuracy ~96% across classes. Precision, recall, F1-score metrics underscore proficiency, with precision values ranging from 0.949 to 0.975 recall 0.957 0.963. comparative analysis state-of-the-art (SOTA) models, outshines counterparts terms precision, F1-score, accuracy. Notably, base architecture, achieves highest values, affirming superiority accurately classifying instances within disease dataset. not only contributes growing body applications but also presents novel effective classification. robust combined competitive edge against SOTA positions it as promising solution crop management.

Язык: Английский

---

Enhancing soybean classification with modified inception model: A transfer learning approach

Emirates Journal of Food and Agriculture, Год журнала: 2024, Номер 36, С. 1 - 9

Опубликована: Апрель 18, 2024

The impact of deep learning (DL) is substantial across numerous domains, particularly in agriculture. Within this context, our study focuses on the classification problematic soybean seeds. dataset employed encompasses five distinct classes, totaling 5513 images. Our model, based InceptionV3 architecture, undergoes modification with addition supplementary layers to enhance efficiency and performance. Techniques such as transfer learning, adaptive rate adjustment (to 0.001), model checkpointing are integrated optimize accuracy. During initial evaluation, achieved 88.07% accuracy training 86.67% validation. Subsequent implementation tuning strategies significantly improves Augmenting architecture additional layers, including Average Pooling, Flatten, Dense, Dropout, Softmax, plays a pivotal role enhancing Evaluation metrics, precision, recall, F1-score, underscore model’s effectiveness. Precision ranges from 0.9706 1.0000, while recall values demonstrate high capture all classes. reflecting balance between precision exhibits remarkable performance ranging 0.9851 1.0000. Comparative analysis existing studies reveals competitive 98.73% by proposed model. While variations exist specific purposes datasets among studies, showcases promising seed classification, contributing advancements agricultural technology for crop health assessment management.

Язык: Английский

---

Computer Vision System for Mango Fruit Defect Detection Using Deep Convolutional Neural Network

Foods, Год журнала: 2022, Номер 11(21), С. 3483 - 3483

Опубликована: Ноя. 2, 2022

Machine learning techniques play a significant role in agricultural applications for computerized grading and quality evaluation of fruits. In the domain, automation improves quality, productivity, economic growth country. The fruits is an essential measure export market, especially defect detection fruit's surface. This pertinent mangoes, which are highly popular India. However, manual mango time-consuming, inconsistent, subjective process. Therefore, computer-assisted system has been developed mangoes. Recently, machine techniques, such as deep method, have used to achieve efficient classification results digital image classification. Specifically, convolution neural network (CNN) technique that employed automated study proposes computer-vision system, employs CNN, After training testing using publicly available database, experimental show proposed method acquired accuracy 98%.

Язык: Английский

---

Generating Image Captions Using Bahdanau Attention Mechanism and Transfer Learning

Symmetry, Год журнала: 2022, Номер 14(12), С. 2681 - 2681

Опубликована: Дек. 18, 2022

Automatic image caption prediction is a challenging task in natural language processing. Most of the researchers have used convolutional neural network as an encoder and decoder. However, accurate requires model to understand semantic relationship that exists between various objects present image. The attention mechanism performs linear combination decoder states. It emphasizes information with visual In this paper, we incorporated Bahdanau two pre-trained networks—Vector Geometry Group InceptionV3—to predict captions given models are encoders Recurrent With help mechanism, able provide context achieve bilingual evaluation understudy score 62.5. Our main goal compare performance on same dataset.

Язык: Английский

---