About the Challenge
Asian Conference on Computer Vision (ACCV) is a leading biennial international conference mainly sponsored by Asian Federation of Computer Vision. This highly successful series provides a premier forum for researchers, developers, and practitioners to present and discuss new problems, solutions, and technologies in computer vision and related areas. Due to the current worldwide COVID-19 pandemic, ACCV 2020 (which was originally scheduled to be held in Kyoto, Japan) will be held as a completely online conference from November 30, 2020 to December 4, 2020.
WebFG 2020 is an international challenge hosted by Nanjing University of Science and Technology, University of Edinburgh, Nanjing University, The University of Adelaide, and Waseda University. It is also supported by Extreme Mart (belonging to Extremevision). This challenge mainly pays attention to the webly-supervised fine-grained recognition problem.
Existing deep learning methods highly depend on large-scale and high-quality labeled training data, which poses a limitation to their practicability and scalability in real world applications. In particular, for fine-grained recognition, a visual task that requires professional knowledge for labeling, the cost of acquiring labeled training data is quite high. So that it is extremely difficult to obtain a large amount of high-quality training data. Therefore, utilizing free web data to train fine-grained recognition models have attracted increasing attentions from researchers in the fine-grained community.
This challenge expects participants to develop webly-supervised fine-grained recognition methods, which leverage web images in training fine-grained recognition models to ease the extreme dependence of deep learning methods on large-scale manually labeled datasets and to enhance their practicability and scalability.
Timeline
October 9th 10:00 AM (UTC+8): Open registration, data download and leaderboard A；
November 27th,10:00 AM (UTC+8): Open leaderboard B；
November 30th, 10:00 AM (UTC+8): Registration deadline；
November 30th, 23:59 PM (UTC+8): Deadline for submission of results；
December 1st, 10:00 AM (UTC+8): Publication of the results of B, Result review;
December 3rd: ACCV 2020 Workshop announced the winners；
Task description
A. Contestants need to register and log in to Extreme Mart platform, fill in the contest registration information;
B. Contestants go through Extreme Mart competition homepage-Data, and download competition training data and test data (Contains List A and List B). Then the Contestants need to conduct algorithm training and development on their local computer or server.
C. After development, contestants need to write the obtained prediction results into a csv file. Then they need to submit the file to Extreme Mart platform.
D. After the prediction result is submitted successfully, the accuracy score and ranking of Contestants on List A will release and the ranking will update once an hour;
E. After 10:00 AM on November 27th (UTC +8)，contestants can choose historical submission results to enter the contest List B. Each team has two chances to submit. Extreme Mart platform will use the highest score among the two submitted results of the contestant as the result of the contestants performance on List B;
F. The final ranking of “List B” will release at 10:00 AM on December 1st (UTC +8). It will be sorted according to the accuracy of the results submitted by the contestants. The final competition ranking is based on the competition List B ;
G. Review: on December 1st (UTC +8), the competition organizing committee will notify the top 3 teams to submit code models and other related documents. They will conduct a manual inspection of these documents. The team that passes the final score review will be the winner of the competition.