Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3
Apples are ranked third, after bananas and oranges, in global fruit production. Fresh apples are more likely to be appreciated by consumers during the marketing process. However, apples inevitably suffer mechanical damage during transport, which can affect their economic performance. Therefore, the...
Main Authors: | , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-04-01
|
Series: | Agriculture |
Subjects: | |
Online Access: | https://www.mdpi.com/2077-0472/13/4/824 |
_version_ | 1797606855442169856 |
---|---|
author | Haiping Si Yunpeng Wang Wenrui Zhao Ming Wang Jiazhen Song Li Wan Zhengdao Song Yujie Li Bacao Fernando Changxia Sun |
author_facet | Haiping Si Yunpeng Wang Wenrui Zhao Ming Wang Jiazhen Song Li Wan Zhengdao Song Yujie Li Bacao Fernando Changxia Sun |
author_sort | Haiping Si |
collection | DOAJ |
description | Apples are ranked third, after bananas and oranges, in global fruit production. Fresh apples are more likely to be appreciated by consumers during the marketing process. However, apples inevitably suffer mechanical damage during transport, which can affect their economic performance. Therefore, the timely detection of apples with surface defects can effectively reduce economic losses. In this paper, we propose an apple surface defect detection method based on weight contrast transfer and the MobileNetV3 model. By means of an acquisition device, a thermal, infrared, and visible apple surface defect dataset is constructed. In addition, a model training strategy for weight contrast transfer is proposed in this paper. The MobileNetV3 model with weight comparison transfer (Weight Compare-MobileNetV3, WC-MobileNetV3) showed a 16% improvement in accuracy, 14.68% improvement in precision, 14.4% improvement in recall, and 15.39% improvement in F1-score. WC-MobileNetV3 compared to MobileNetV3 with fine-tuning improved accuracy by 2.4%, precision by 2.67%, recall by 2.42% and F1-score by 2.56% compared to the classical neural networks AlexNet, ResNet50, DenseNet169, and EfficientNetV2. The experimental results show that the WC-MobileNetV3 model adequately balances accuracy and detection time and achieves better performance. In summary, the proposed method achieves high accuracy for apple surface defect detection and can meet the demand of online apple grading. |
first_indexed | 2024-03-11T05:20:52Z |
format | Article |
id | doaj.art-67b82c85f3cc40a3b523ad3bed1ab779 |
institution | Directory Open Access Journal |
issn | 2077-0472 |
language | English |
last_indexed | 2024-03-11T05:20:52Z |
publishDate | 2023-04-01 |
publisher | MDPI AG |
record_format | Article |
series | Agriculture |
spelling | doaj.art-67b82c85f3cc40a3b523ad3bed1ab7792023-11-17T17:53:59ZengMDPI AGAgriculture2077-04722023-04-0113482410.3390/agriculture13040824Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3Haiping Si0Yunpeng Wang1Wenrui Zhao2Ming Wang3Jiazhen Song4Li Wan5Zhengdao Song6Yujie Li7Bacao Fernando8Changxia Sun9College of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaNOVA Information Management School (NOVA IMS), Universidade Nova de Lisboa, 1099-085 Lisbon, PortugalCollege of Information and Management Science, Henan Agricultural University, Zhengzhou 450046, ChinaApples are ranked third, after bananas and oranges, in global fruit production. Fresh apples are more likely to be appreciated by consumers during the marketing process. However, apples inevitably suffer mechanical damage during transport, which can affect their economic performance. Therefore, the timely detection of apples with surface defects can effectively reduce economic losses. In this paper, we propose an apple surface defect detection method based on weight contrast transfer and the MobileNetV3 model. By means of an acquisition device, a thermal, infrared, and visible apple surface defect dataset is constructed. In addition, a model training strategy for weight contrast transfer is proposed in this paper. The MobileNetV3 model with weight comparison transfer (Weight Compare-MobileNetV3, WC-MobileNetV3) showed a 16% improvement in accuracy, 14.68% improvement in precision, 14.4% improvement in recall, and 15.39% improvement in F1-score. WC-MobileNetV3 compared to MobileNetV3 with fine-tuning improved accuracy by 2.4%, precision by 2.67%, recall by 2.42% and F1-score by 2.56% compared to the classical neural networks AlexNet, ResNet50, DenseNet169, and EfficientNetV2. The experimental results show that the WC-MobileNetV3 model adequately balances accuracy and detection time and achieves better performance. In summary, the proposed method achieves high accuracy for apple surface defect detection and can meet the demand of online apple grading.https://www.mdpi.com/2077-0472/13/4/824defect detectionimage fusiondeep learningtransfer learningweight comparison |
spellingShingle | Haiping Si Yunpeng Wang Wenrui Zhao Ming Wang Jiazhen Song Li Wan Zhengdao Song Yujie Li Bacao Fernando Changxia Sun Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 Agriculture defect detection image fusion deep learning transfer learning weight comparison |
title | Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 |
title_full | Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 |
title_fullStr | Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 |
title_full_unstemmed | Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 |
title_short | Apple Surface Defect Detection Method Based on Weight Comparison Transfer Learning with MobileNetV3 |
title_sort | apple surface defect detection method based on weight comparison transfer learning with mobilenetv3 |
topic | defect detection image fusion deep learning transfer learning weight comparison |
url | https://www.mdpi.com/2077-0472/13/4/824 |
work_keys_str_mv | AT haipingsi applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT yunpengwang applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT wenruizhao applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT mingwang applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT jiazhensong applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT liwan applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT zhengdaosong applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT yujieli applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT bacaofernando applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 AT changxiasun applesurfacedefectdetectionmethodbasedonweightcomparisontransferlearningwithmobilenetv3 |