Single Stage Deep Transfer Learning Model for Apparel Detection and Classification for E-Commerce

Ssvr Kumar Addagarla ； Anthoniraj Amalanathan

《International Journal of Electronic Commerce Studies》 13卷1期 (2022/03) Pp. 69-92

https://doi.org/10.7903/ijecs.1953

摘要

Although many computer vision-based object detection techniques have evolved in the past decade, it suffers from inconsistent detection accuracy, especially for multiclass classification problems. This paper proposed an approach using the Single Stage Deep Transfer Learning model (SS-DTLM) for multiclass apparel detection using a customized YoloV3 algorithm by adapting 3-level Spatial pyramid pooling (SPP), a multi-scale image feature extractor for faster and reasonable apparel detection and classification. This approach produced a reasonable Mean Average Precision (mAP), reliable object detection, and classification. Our model trained and tested on Open Images Dataset (OIDV4) with six object classes and Custom built Apparel Dataset with five object classes of apparels. Finally, experimental results compared with baseline Yolov3 and Yolov3-Tiny algorithms. Further, this paper also emphasized the detected image's various color spaces using SS-DTLM by applying the K-Means clustering algorithm for further analysis.

關鍵字

Custom Object Detection ； Yolov3 ； Spatial pyramid pooling ； Color Space ； Apparel detection

參考文獻

P. Sinha, B. Balas, Y. Ostrovsky, and R. Russell, “Face recognition by humans: Nineteen results all computer vision researchers should know about,” Proc. IEEE, vol. 94, no. 11, pp. 1948–1962, 2006. https://doi.org/10.1109/JPROC.2006.884093

S. Zoghbi, G. Heyman, J. C. Gomez, and M.-F. Moens, “Fashion meets computer vision and nlp at e-commerce search,” Int. J. Comput. Electr. Eng., vol. 8, no. 1, pp. 31–43, 2016. https://doi.org/10.17706/IJCEE.2016.8.1.31-43

K. Hara, V. Jagadeesh, and R. Piramuthu, “Fashion apparel detection: The role of deep convolutional neural network and pose-dependent priors,” in 2016 IEEE Winter Conference on Applications of Computer Vision (WACV), 2016, pp. 1–9. https://doi.org/10.1109/WACV.2016.7477611

N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” in 2005 IEEE computer society conference on computer vision and pattern recognition (CVPR’05), 2005, vol. 1, pp. 886–893. https://doi.org/10.1109/CVPR.2005.177

X. Wang, T. X. Han, and S. Yan, “An HOG-LBP human detector with partial occlusion handling,” in 2009 IEEE 12th international conference on computer vision, 2009, pp. 32–39. https://doi.org/10.1109/ICCV.2009.5459207