Self-supervised Video Representation Learning with Cross-Stream Prototypical Contrasting

Open paper
Benchmark Model Rank Results
self-supervised-action-recognition-on-hmdb51ViCC (S3D; R+F)#22Top-1 Accuracy: 62.2Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-hmdb51ViCC (R2+1D; R+F)#23Top-1 Accuracy: 61.5Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-hmdb51ViCC (R2+1D; RGB)#30Top-1 Accuracy: 52.4Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-hmdb51ViCC (S3D; RGB)#33Top-1 Accuracy: 38.5Pre-Training Dataset: UCF101Frozen: true
self-supervised-action-recognition-on-hmdb51-1ViCC (S3D; R+F)#8Top-1 Accuracy: 62.2Pretraining Dataset: UCF101
self-supervised-action-recognition-on-hmdb51-1ViCC (R2+1D; RGB)#10Top-1 Accuracy: 52.4Pretraining Dataset: UCF101
self-supervised-action-recognition-on-hmdb51-1ViCC (S3D; RGB))#11Top-1 Accuracy: 47.9Pretraining Dataset: UCF101
self-supervised-action-recognition-on-ucf101ViCC (S3D; R+F)#213-fold Accuracy: 90.5Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-ucf101ViCC (S3D; RGB)#223-fold Accuracy: 88.8Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-ucf101ViCC (R2+1D; R+F)#233-fold Accuracy: 88.8Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-ucf101ViCC (R2+1D; RGB)#273-fold Accuracy: 82.8Pre-Training Dataset: UCF101Frozen: false
self-supervised-action-recognition-on-ucf101ViCC (S3D; RGB)#333-fold Accuracy: 72.2Pre-Training Dataset: UCF101Frozen: true
self-supervised-action-recognition-on-ucf101-1ViCC (S3D; R+F)#83-fold Accuracy: 90.5Pretrain: UCF101
self-supervised-action-recognition-on-ucf101-1ViCC (R2+1D; R+F)#93-fold Accuracy: 88.8Pretrain: UCF101
self-supervised-action-recognition-on-ucf101-1ViCC (S3D; RGB)#113-fold Accuracy: 84.3Pretrain: UCF101
self-supervised-action-recognition-on-ucf101-1ViCC (R2+1D; RGB)#123-fold Accuracy: 82.8Pretrain: UCF101

Rank counts only results with a code link.