Leveraging Temporal Contextualization for Video Action Recognition

Benchmark Model Rank Results
zero-shot-action-recognition-on-hmdb51TC-CLIP#4Top-1 Accuracy: 56.0
zero-shot-action-recognition-on-kineticsTC-CLIP#1Top-1 Accuracy: 78.1Top-5 Accuracy: 95.7
zero-shot-action-recognition-on-ucf101TC-CLIP#4Top-1 Accuracy: 85.4