Can we build visual evaluation systems with/for large-scale generative models (LGMs)? Challenge Challenge Tracks The general focus of the challenges is to measure how emerging visual evaluation systems can precisely answer open-ended questions related to the corresponding 4 track scopes. To ensure more objective quantitative evaluation among participants, all track questions are multi-choice questions (MCQ) with multiple candidate options provided, and final metrics will be based on the accuracy on answerin
[September 2022] Description of the news.
For more details, please visit the detailed documentation .