VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs

Benchmark Model Rank Results
audio-classification-on-vocalsoundVideoLLaMA 2-AV 7B#2Accuracy: 93.19
video-question-answering-on-perception-testVideoLLaMA2 (72B)#4Accuracy (Top-1): 57.5
temporal-relation-extraction-on-vinogroundVideoLLaMA2-72B#6Text Score: 36.2Video Score: 21.8Group Score: 8.4
zero-shot-video-question-answer-on-vnbenchVideoLLaMA2#8Accuracy: 4.5
zero-shot-video-question-answer-on-video-mmeVideoLLaMA2 (72B)#9Accuracy (%): 63.1
video-question-answering-on-tvbenchVideoLLaMA2 72B#10Average Accuracy: 48.4
video-question-answering-on-tvbenchVideoLLaMA2 7B#14Average Accuracy: 42.9
video-question-answering-on-tvbenchVideoLLaMA2.1#17Average Accuracy: 42.1
zeroshot-video-question-answer-on-msvd-qaVideoLLaMA 2 7B#16Accuracy: 71.7
zero-shot-video-question-answer-on-egoschema-fullsetVideoLLaMA2 (72B)#18Accuracy: 63.9
video-question-answering-on-next-qaVideoLLaMA2.1(7B)#22Accuracy: 75.6
zero-shot-video-question-answer-on-video-mme-w-o-subsVideoLLaMA2 (72B)#48Accuracy (%): 60.9
video-question-answering-on-mlvuVideoLLaMA 2 72B#59Accuracy: 45.6
video-question-answering-on-mvbenchVideoLLaMA2 (72B)#62Avg.: 62.0