패션 카테고리 오버샘플링 자동화 시스템
Fashion Category Oversampling Automation System
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
국내 온라인 패션 플랫폼은 개인사업자가 제품정보를 직접 등록하기 때문에 개인사업자의 불편함을 초래한다. 많은 제품군을 한꺼번에 수동 등록하므로 수기 입력된 제품정보로 인한 신뢰성문제가 발생한다. 등록된 상품 이미지의 저품질 및 데이터 수의 불균형으로 인한 편향도 심각하게 제기된다. 본 연구는 오버샘플링 기법을 통해 데이터 편향을 최소화하고 13개 패션 카테고리의 다중 분류를 수행하는 ResNet50 모델을 제안한다. 컴퓨팅 자원과 오랜 학습시간을 최소화하기위해 전이학습을 활용했다. 결과적으로, 데이터 수가 매우 부족했던 클래스의 데이터 증강을 통해기본 CNN 모델에 비해 최대 33.4%의 향상된 식별력을 보여주었다. 모든 결과의 신뢰성은 정밀도-재현율 곡선으로 보장한다. 본 연구는 국내 온라인 패션 플랫폼 산업의 발전을 한 단계 끌어올릴수 있을 것으로 기대한다.
Abstract
In the realm of domestic online fashion platform industry the manual registration of product information by individual business owners leads to inconvenience and reliability issues, especially when dealing with simultaneous registrations of numerous product groups. Moreover, bias is significantly heightened due to the low quality of product images and an imbalance in data quantity. Therefore, this study proposes a ResNet50 model aimed at minimizing data bias through oversampling techniques and conducting multiple classifications for 13 fashion categories. Transfer learning is employed to optimize resource utilization and reduce prolonged learning times. The results indicate improved discrimination of up to 33.4% for data augmentation in classes with insufficient data compared to the basic convolution neural network (CNN) model. The reliability of all outcomes is underscored by precision and affirmed by the recall curve. This study is suggested to advance the development of the domestic online fashion platform industry to a higher echelon.
참고문헌 (18)
- Sen Bin, "Home Fashion in the COVID-19 Era Identified through Big Data Analysis", Journal of the Korean Society of Costume, 72(5), pp.92~108 (2022)[DOI]
- Agrawal, Durgesh Kumar, "Increasing digital dissemination and online apparel shopping behaviour of Generation Y", Journal of Fashion Marketing and Management (2023)[DOI]
- Koch, J., "Online shopping motives during the COVID-19 pandemic—lessons from the crisis", Sustainability, 12(24), pp.10247~ (2020)[DOI]
- "The size of the Korean fashion market in 2023has been trending for five years"[URL]
- "KRW 45.7787 trillion in the domestic fashion market in 2022"[URL]
- L. Kessous, "Multimodal emotion recognition in speech-based interaction using facial expression, body gesture and acoustic analysis", J. Multimodal User Interfaces, 3(1-2), pp.33~48 (2010)[DOI]
- "Musinsa tops visitors to online fashion malls in Korea"[URL]
- "Apier Announces Successful Cases of online fashion platform ‘Mushinsa’ App"[URL]
- "It shakes the market of NAVER, Shopify, and e-commerce with stable growth through exposure to ‘store’ rather than ‘product’"[URL]
- "Musinsa"[URL]
- "Frequently asked questions regarding Naver Shopping's entry and return"[URL]
- Liu, Z., "Deepfashion : Powering robust clothes recognition and retrieval with rich annotations", Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), pp.1096~1104 (2016)
- Ge, Y., "Deepfashion2 : A versatile benchmark for detection, pose estimation, segmentation and re-identification of clothing images", Proceedings of the IEEE, 2019, 5337~5345
- Kim, I., "Metadata Design and Machine Learning-Based Automatic Indexing for Efficient Data Management of Image Archives of Local Governments in South Korea", Journal of Korean Society of Archives and Records Management, 20(2), pp.67~83 (2020)[DOI]
- "DeepFashion2"[URL]
- MAJURAN, Shajini, "A single-stage fashion clothing detection using multilevel visual attention", The Visual Computer, 39, pp.6609~6623 (2022)[DOI]
- Zhao, L., "Neo-fashion : A data-driven fashion trend forecasting system using catwalk analysis", Clothing and Textiles Research Journal, 42(1) (2021)[DOI]
- Chakraborty, S., "Predicting fashion trend using runway images : application of logistic regression in trend forecasting", International Journal of Fashion Design, Technology and Education, 13(3), pp.376~386 (2020)[DOI]