HyperAIHyperAI

Command Palette

Search for a command to run...

Console

Action Recognition On Diving 48

Metrics

Accuracy

Results

Performance results of various models on this benchmark

Paper Title
LVMAE94.9Extending Video Masked Autoencoders to 128 frames
Video-FocalNet-B90.8Video-FocalNets: Spatio-Temporal Focal Modulation for Video Action Recognition
AIM (CLIP ViT-L/14, 32x224)90.6AIM: Adapting Image Models for Efficient Video Action Recognition
DUALPATH88.7Dual-path Adaptation from Image to Video Transformers
StructVit-B-4-188.3Learning Correlation Structures for Vision Transformers
TFCNet88.3TFCNet: Temporal Fully Connected Networks for Static Unbiased Temporal Reasoning
ORViT TimeSformer88.0Object-Region Video Transformers
GC-TDN87.6Group Contextualization for Video Recognition
BEVT86.7BEVT: BERT Pretraining of Video Transformers
PSB86Spatiotemporal Self-attention Modeling with Temporal Patch Shift for Action Recognition
VIMPAC85.5VIMPAC: Video Pre-Training via Masked Token Prediction and Contrastive Learning
RSANet-R50 (16 frames, ImageNet pretrained, a single clip)84.2Relational Self-Attention: What's Missing in Attention for Video Understanding
TQN81.8Temporal Query Networks for Fine-grained Video Understanding
PMI Sampler81.3PMI Sampler: Patch Similarity Guided Frame Selection for Aerial Action Recognition
TimeSformer-L81Is Space-Time Attention All You Need for Video Understanding?
TimeSformer-HR78Is Space-Time Attention All You Need for Video Understanding?
SlowFast77.6SlowFast Networks for Video Recognition
TimeSformer75Is Space-Time Attention All You Need for Video Understanding?
0 of 18 row(s) selected.
Action Recognition On Diving 48 | SOTA | HyperAI