HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection
Accurate inspection of subtle defects on apple surfaces is necessary in agricultural engineering. However, existing methods often rely on expensive equipment and encounter difficulty in detecting small defect areas effectively. To address this challenge, we introduce the Subtle Surface Defects in Ap...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10776948/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850264699134803968 |
|---|---|
| author | Nguyen Bui Ngoc Han Ju-Hwan Lee Dang Thanh Vu Iqbal Murtza Hyoung-Gook Kim Jin-Young Kim |
| author_facet | Nguyen Bui Ngoc Han Ju-Hwan Lee Dang Thanh Vu Iqbal Murtza Hyoung-Gook Kim Jin-Young Kim |
| author_sort | Nguyen Bui Ngoc Han |
| collection | DOAJ |
| description | Accurate inspection of subtle defects on apple surfaces is necessary in agricultural engineering. However, existing methods often rely on expensive equipment and encounter difficulty in detecting small defect areas effectively. To address this challenge, we introduce the Subtle Surface Defects in Apples (SSDA) dataset, a custom dataset specifically collected and annotated for two defect types: scratches and pest damage. For benchmarking, we propose a heatmap-based anchor-free (HAFREE) detector, a novel end-to-end object detection architecture designed to localize subtle defects on apple surfaces. Unlike prior methods that rely on anchor boxes, HAFREE employs a heatmap-based approach to represent defects as keypoints using two-dimensional Gaussian heatmaps. We introduce a multiscale feature fusion block, targeting small objects by incorporating local and global contextual information. To mitigate overfitting, we also implement a patch training strategy incorporating full images and cropped patches during training as a regularizer. The proposed method achieves a mAP50 of 50.05% on the SSDA dataset, outperforming one- and two-stage anchor-based detectors and previous anchor-free approaches. Code is available at: <uri>https://github.com/nbngochan/HAFREE</uri>. |
| format | Article |
| id | doaj-art-8abd5172e9ec4296b261c076a0a5dd4b |
| institution | OA Journals |
| issn | 2169-3536 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-8abd5172e9ec4296b261c076a0a5dd4b2025-08-20T01:54:38ZengIEEEIEEE Access2169-35362024-01-011218279918281310.1109/ACCESS.2024.350439910776948HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect DetectionNguyen Bui Ngoc Han0https://orcid.org/0009-0007-7462-662XJu-Hwan Lee1https://orcid.org/0000-0002-1553-9637Dang Thanh Vu2https://orcid.org/0000-0003-0743-8251Iqbal Murtza3https://orcid.org/0000-0001-7541-8581Hyoung-Gook Kim4https://orcid.org/0000-0002-1518-4100Jin-Young Kim5https://orcid.org/0000-0002-4896-8980Department of Electronic Convergence Engineering, Kwangwoon University, Seoul, South KoreaDepartment of Intelligent Electronics and Computer Engineering, Chonnam National University, Gwangju, South KoreaResearch Center, AISeed Inc., Gwangju, South KoreaDepartment of Intelligent Electronics and Computer Engineering, Chonnam National University, Gwangju, South KoreaDepartment of Electronic Convergence Engineering, Kwangwoon University, Seoul, South KoreaDepartment of Electronic Convergence Engineering, Kwangwoon University, Seoul, South KoreaAccurate inspection of subtle defects on apple surfaces is necessary in agricultural engineering. However, existing methods often rely on expensive equipment and encounter difficulty in detecting small defect areas effectively. To address this challenge, we introduce the Subtle Surface Defects in Apples (SSDA) dataset, a custom dataset specifically collected and annotated for two defect types: scratches and pest damage. For benchmarking, we propose a heatmap-based anchor-free (HAFREE) detector, a novel end-to-end object detection architecture designed to localize subtle defects on apple surfaces. Unlike prior methods that rely on anchor boxes, HAFREE employs a heatmap-based approach to represent defects as keypoints using two-dimensional Gaussian heatmaps. We introduce a multiscale feature fusion block, targeting small objects by incorporating local and global contextual information. To mitigate overfitting, we also implement a patch training strategy incorporating full images and cropped patches during training as a regularizer. The proposed method achieves a mAP50 of 50.05% on the SSDA dataset, outperforming one- and two-stage anchor-based detectors and previous anchor-free approaches. Code is available at: <uri>https://github.com/nbngochan/HAFREE</uri>.https://ieeexplore.ieee.org/document/10776948/Anchor-free detectordeep learningfruit datasetGaussian mapobject detectionsubtle defect detection |
| spellingShingle | Nguyen Bui Ngoc Han Ju-Hwan Lee Dang Thanh Vu Iqbal Murtza Hyoung-Gook Kim Jin-Young Kim HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection IEEE Access Anchor-free detector deep learning fruit dataset Gaussian map object detection subtle defect detection |
| title | HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection |
| title_full | HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection |
| title_fullStr | HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection |
| title_full_unstemmed | HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection |
| title_short | HAFREE: A Heatmap-Based Anchor-Free Detector for Apple Defect Detection |
| title_sort | hafree a heatmap based anchor free detector for apple defect detection |
| topic | Anchor-free detector deep learning fruit dataset Gaussian map object detection subtle defect detection |
| url | https://ieeexplore.ieee.org/document/10776948/ |
| work_keys_str_mv | AT nguyenbuingochan hafreeaheatmapbasedanchorfreedetectorforappledefectdetection AT juhwanlee hafreeaheatmapbasedanchorfreedetectorforappledefectdetection AT dangthanhvu hafreeaheatmapbasedanchorfreedetectorforappledefectdetection AT iqbalmurtza hafreeaheatmapbasedanchorfreedetectorforappledefectdetection AT hyounggookkim hafreeaheatmapbasedanchorfreedetectorforappledefectdetection AT jinyoungkim hafreeaheatmapbasedanchorfreedetectorforappledefectdetection |