VideoTGB icon indicating copy to clipboard operation
VideoTGB copied to clipboard

[EMNLP 2024] A Video Chat Agent with Temporal Prior

Results 6 VideoTGB issues
Sort by recently updated
recently updated
newest added

作者您好,我注意到您的工作中使用了光流模态,我尝试从长视频中提取光流,但是其耗时较长,存储成本较高。想请问您是否对长视频作了特殊的处理,谢谢 -v-

in demo.py model_base = 'ckpts/instructblip-vicuna-7b' sampler_model_base = 'ckpts/bert-base-uncased' Are these two weight files downloaded from https://huggingface.co/Salesforce/instructblip-vicuna-7b and https://huggingface.co/google-bert/bert-base-uncased?

Have you processed the original question files and output the new test_q.json and test_a.json files? If yes, could you please provide the test_q.json and test_a.json files? Or the process.py file...

作者,您好!您可以提供光流和伪标签吗?

您好,作者,又打扰了! 能不能请您提供VideoTGB/inputs/videoinstruct这个文件夹下面的所有文件呀? 尝试运行代码的过程中,仿佛永远都在需要新的文件。 先是FileNotFoundError: [Errno 2] No such file or directory: '/root/VideoTGB/inputs/videoinstruct/flow',等我解决了这个问题, 接着是FileNotFoundError: [Errno 2] No such file or directory: '/root/VideoTGB/inputs/videoinstruct/train.json', 接着是FileNotFoundError: [Errno 2] No such file or directory: '/root/VideoTGB/inputs/videoinstruct/pesudo_label.json'...

Hello, thank you for your excellent work. I would like to ask, how can I generate pseudo-labels? Could you provide the code for generating pseudo-labels?