- General Chairs:
- Yueting Zhuang,
- Xing Tang,
- Guilin Wu,
- Yahong Han,
- Haihong Tang,
- Xiaobo Li,
- Xiaohan Wang,
- Baoming Yan,
- Bo Gao,
- Yi Yang
It is our great pleasure to welcome you to the 1st Workshop on Multimodal Product Identification in Livestreaming and WAB Challenge - WAB 2021, which is held in conjunction with the ACM Multimedia 2021. The WAB Challenge and associated workshop is held to allow researchers to present their progress, communicate and co-develop novel ideas that potentially shape the future of multimodal retrieval. Based on the proposed multimodal product retrieval dataset named "Watch and Buy" (WAB), we launch WAB Challenge, which encourages participants to establish approaches to fully automatic detection and multimodal identification of products presented in real-world e-commerce livestreaming.
The challenge attracted registrations of 587 teams from Asia, Europe, and North America. After preliminaries, the top 20 teams enroll in the semi-finals. After semi-finals, the top 10 teams enroll in the finals. Then based on end2end reproduced results and technical solution report, the final leaderboard was determined. The best model achieves the averaged F1 score of 0.69, which is 0.22 higher than the baseline model. All of the top-3 teams have a higher score than 0.6. For the call for papers part, the program committee accepted 5 papers, including 2 long papers and 3 short papers.
Proceeding Downloads
Domain Balanced Sampling and Iterative Search for Product Identification
This paper introduces our solution to the 1st Workshop on Multimodal Product Identification in Livestreaming and Watch and Buy Challenge, a real-world task in a live stream scene and is challenging due to factors such as lighting, occlusion, and cross-...
Multimodal Product Identification: Submission to Watch and Buy 2021 Challenge
This technical report describes the overview of our approach to the "Watch and Buy: Multimodal Product Identification Challenge". Specifically, we tackle this problem with a three-stage framework, i.e., product detection, retrieval and classification. ...
3rd Solution to WAB Challenge: A Two Stage Model Using only Pixel-level Features for Video-Image Retrieval
In this article, we propose an effective product retrieval system pipeline robust to large-scale E-commerce platform product data. Our proposed method consists of three parts: detection, retrieval, and post-processing. We first perform the detection ...
Multimodal Region-level Clothing Re-identification in E-commerce Livestreaming
This is a technical report on the 1st Workshop on Multimodal Product Identification in Livestreaming and WAB Challenge. The technical pipeline includes, 1)object detection. 2)Product Reidentification. 3)Multimodal classification under multi-criteria ...
Watch and Buy: A Practical Solution for Real-time Fashion Product Identification in Live Stream
"Watch and Buy: Multimodal Product Identification(WAB)" challenge is a new task in the field of cross-modal retrieval, which aims to retrieve the relevant products when users watching live streamers selling fashion products. In practice, it is very hard ...
Cited By
- Gu S, Yuan X, Zhao L, Liu Z, Hu Y and Chen Z MVCIR-net: Multi-view Clustering Information Reinforcement Network Proceedings of the 31st ACM International Conference on Multimedia, (3609-3618)
- Li B, Zheng H, Zhang Z, Zhao Y, Zhao Z and Zhang H Dynamic Grouped Interaction Network for Low-Light Stereo Image Enhancement Proceedings of the 31st ACM International Conference on Multimedia, (2468-2476)
Recommendations
WAB'21: 1st Workshop on Multimodal Product Identification in Livestreaming and WAB Challenge
MM '21: Proceedings of the 29th ACM International Conference on MultimediaProduct identification has become a very important component in the modern E-commerce shopping system. Consumers could enjoy watching livingstreaming and buying products that livestream hosts recommended. However, with hundreds of products presented in ...