MAXI

Reported on 4 benchmarks across 1 task · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision4 results

Zero-Shot Action RecognitiononKinetics
Top-1 Accuracy· 2023-03-15
71.6
best: 78.1 (TC-CLIP)
SOTA
MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge arXiv:2303.08914
Zero-Shot Action RecognitiononUCF101
Top-1 Accuracy· 2023-03-15
78.2
best: 92.8 (OTI(ViT-L/14))
MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge arXiv:2303.08914
Zero-Shot Action RecognitiononCharades
mAP· 2023-03-15
23.8
best: 35.59 (MSQNet)
MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge arXiv:2303.08914
Zero-Shot Action RecognitiononHMDB51
Top-1 Accuracy· 2023-03-15
52.3
best: 64.7 (MOV (ViT-L/14))
MAtch, eXpand and Improve: Unsupervised Finetuning for Zero-Shot Action Recognition with Language Knowledge arXiv:2303.08914