pku-yuangroup / video-bench Goto Github PK

Hi, thanks for sharing the impressive benchmark.

I'm particularly interested in understanding more about the multi-choice questions used for assessing Basic QA ability. Could you kindly provide additional information on how they are obtained? (generated by ChatGPT or manually written?)

The provided generic step1 code throws error about 'answer' key does not exists in the provided EVALQA json files.

Hi, I want to evaluate my model result on UCF-CRIME dataset, and I made the json file Video-Bench-Input.json. When I press the "submit eval" botton on the hugging face page of Video-Bench Leaderboard, it seems no response. And I clicked ‘Refresh’ to obtain the uploaded leaderboard, no uploaded yet. What's the problem? Please help me to solve it, thanks a lot!

the real eval is done by your website, right?

I see some impressive results of Sphinx-V2 on the leaderboard. How to process the video data using Sphinx-V2?

Hi.

I noticed that there is no code that will give me the score before submitting to the leaderboard.
Want to check my scores before submitting it and if it's low, I'll work on it more and resubmit it maybe.

Do you have plans for releasing the video_bench scoring code?