Too Many Frames, Not All Useful: Efficient Strategies for Long-Form Video QA

Benchmark Model Rank Results
zero-shot-video-question-answer-on-intentqaLVNet#1Accuracy: 71.1
zero-shot-video-question-answer-on-egoschema-subsetLVNet#5Accuracy: 66.0
zero-shot-video-question-answer-on-next-qaLVNet(GPT-4o)#8Accuracy: 72.9
zero-shot-video-question-answer-on-egoschema-fullsetLVNet#28Accuracy: 61.1