Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models

Benchmark Model Rank Results
question-answering-on-next-qa-open-endedVideo-ChatGPT#4Accuracy: 54.6Confidence Score: 3.2
vcgbench-diverse-on-videoinstructVideo-ChatGPT#6mean: 2.08Correctness of Information: 2.07Detail Orientation: 2.42
video-based-generative-performanceVideo-ChatGPT#20mean: 2.38Correctness of Information: 2.4Detail Orientation: 2.52
video-based-generative-performance-1Video-ChatGPT#14gpt-score: 2.40
video-based-generative-performance-2Video-ChatGPT#14gpt-score: 2.37
video-based-generative-performance-3Video-ChatGPT#15gpt-score: 2.62
video-based-generative-performance-4Video-ChatGPT#14gpt-score: 2.52
video-based-generative-performance-5Video-ChatGPT#16gpt-score: 1.98
video-question-answering-on-activitynet-qaVideo-ChatGPT#25Accuracy: 35.2Confidence score: 2.7
video-question-answering-on-mvbenchVideo-ChatGPT#20Avg.: 32.7
zero-shot-video-question-answer-on-vnbenchVideoChatGPT#9Accuracy: 4.1
zeroshot-video-question-answer-on-activitynetVideo-ChatGPT#24Accuracy: 35.2Confidence Score: 2.7
zeroshot-video-question-answer-on-msrvtt-qaVideo-ChatGPT-7B#26Accuracy: 49.3Confidence Score: 2.8
zeroshot-video-question-answer-on-msvd-qaVideo-ChatGPT-7B#24Accuracy: 64.9Confidence Score: 3.3
zeroshot-video-question-answer-on-tgif-qaVideo-ChatGPT-7B#12Accuracy: 51.4Confidence Score: 3.0