1.

Record Nr.

UNINA9910510426603321

Titolo

WAB '21 : proceedings of the 1st Workshop on Multimodal Product Identification in Livestreaming and WAB Challenge : October 24, 2021, Virtual Event, China / / Yueting Zhuang [and nine others], chairs

Pubbl/distr/stampa

New York : , : The Association for Computing Machinery, , 2021

Descrizione fisica

1 online resource (xi, 33 pages) : illustrations

Collana

ACM international conference proceedings series

Disciplina

025.04

Soggetti

Information retrieval

Manufactures

Multimodal user interfaces (Computer systems)

Lingua di pubblicazione

Inglese

Formato

Materiale a stampa

Livello bibliografico

Monografia

Note generali

Includes index.

Sommario/riassunto

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.