CC BY
20For citation-. Shi Yishan. Method on Fabric defect detection based on deep learning //
URL: http.//rectors.altstu.ru/ru/periodical/archiv/2022/2/articles/2_4.pdf EDN. https://elibrary.ru/moqqbn
UDK 677.019
Method on Fabric defect detection based on deep learning
1 «IfS^ê^m 430073;
2 ^n 431700 E-mail: [email protected]
0
immn,
, fê^TU R-CNN [1] (region-convolution neural network), Fast-RCNN [2], Faster-RCNN [3]
s ^m^^i^ram^j s . u
SSD (single shot multibox detector) [4], YOLO (you only look once) [5], YOLOV2 [6], YOLOV3 [7], YOLOV6 ^
uster
Fabriscan g^M«^, tt^Bt Barco ^Élg^M«^, fêS Obdix ^Wfê^^ai^m^m^, UÉ^J EVS I-Tex ê
[8] »Mi^M^M^, Mfê^, #AAIWTAI$*.
^fflftfêffl Faster-RCNN SPM^HfêlWè, Xtii
«7 1500 ШШйШ,Мй^Щ,
тижшгшшшш. шшшш®, m шш, ш, ет, ш
Щ, ШШШЯ 230 512x512, 1 ЖЖ.
(a)
(b)
ffl1
Fig. 1. Sample pictures of defects
(c) Mâ
m, , wmmmrnmfrn^, ж«^,
mm, шщшгшммшш, шёшммшш 7:2:1 é^m^ju
ШШ, ШШтММКШ.
/S
шштшттттшл
^ХШ ResNet50 ШШШШШ, Ш ReLU m
Batch Normalization
КЖШШШШ^ТШ. «ЯЖШШЖШ Faster RCNN ШШ
ш, ш 2 тж.
ffl 2 Faster RCNN IN^M
Fig. 2. Structure of the improved Faster RCNN network
ад, ш
ЛResNet50 m FPN + Faster RCNN Щ
a 3 тж.
,
DoiSho 0.8В» •»--*•• • - ...
шв
(а)
(b) ЙЙ
ш з т&тття^
Fig. 3. Partial results of defect detection
(c) МЙ
ResNet50 ШШШММ^ 99.28%, ШШМ^ 98.09%, ШШ
85.93%. ШШШШШШ& Faster RCNN ЖШ, МШ&Й, 3 ^^
ттшш, ^шм^л^т ^мте^
#Ш. ^ХШШ Pytorch ШШШ Faster RCNN ЖШШШ,
мм, Ж'шштъш, шшш
ШЖШ, ШШШШШ, ^Ш^ШПШ-^ШЯ.
[1] Girshick R., Donahue J., Darrell T.et al. Rich feature hierarchies for accurate object detection and semantic segmentation [C] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2014:580-587.
[2] Girshick R. Fast r-cc[C] // Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2015:1440-1448.
[3] Ren S., He K., Girshick R. et al. Faster r-cnn: towards real-time object detection with region proposal networks [C] // IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(6):1137-1149.
[4] Liu Wei, et al. SSD: single shot multibox detector[C] // European Conference on Computer Vision, 2016:21-37.
[5] Redmon Joseph, et al. You only look once: Unified, real-time object detection [C] // Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2016:779-788.
[6] Redmon Joseph, et al. Y0L09000: better, faster, stronger [C] // Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2017:7263-7271.
[7] Redmon, Joseph, et al. Yolov3: an incremental improvement [C] // Computer Vision and Pattern Recognition, arXiv preprint arXiv:1804.02767,2018.
[8] [J]. 2014(03)
References
1. Girshick R., Donahue J., Darrell T., et al. Rich feature hierarchies for accurate object detection and semantic segmentation [C] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2014:580-587.
[2] Girshick R. Fast r-cc[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2015:1440-1448.
[3] Ren S., He K., Girshick R., et al. Faster r-cnn: towards real-time object detection with region proposal networks[C]//IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(6):1137-1149.
[4] Liu, Wei, et al. SSD: single shot multibox detector[C]//European Conference on Computer Vision, 2016:21-37.
[5] Redmon, Joseph, et al. You only look once: Unified, real-time object detection [C] // Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2016:779-788.
[6] Redmon, Joseph, et al. Y0L09000: better, faster, stronger [C] // Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2017:7263-7271.
[7] Redmon, Joseph, et al. Yolov3: an incremental improvement [C] //Computer Vision and Pattern Recognition, arXiv preprint arXiv:1804.02767,2018.
[8] Li Wenyu, Cheng Longdi. Fabric Defect Detection Based on Machine Vision and image Processing [J]. Journal of Textile Science and Technology, 2014(03)
