Action Recognition In Videos On Ucf101

评估指标

3-fold Accuracy

评测结果

各个模型在此基准测试上的表现结果

Paper TitleRepository
FTP-UniFormerV2-L/1499.7Enhancing Video Transformers for Action Understanding with VLM-aided Training-
OmniVec299.6OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning-
OmniVec99.6OmniVec: Learning robust representations with cross modal sharing-
VideoMAE V2-g99.6VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking
BIKE98.8Bidirectional Cross-Modal Knowledge Exploration for Video Recognition with Pre-trained Vision-Language Models
SMART98.64SMART Frame Selection for Action Recognition-
OmniSource (SlowOnly-8x8-R101-RGB + I3D-Flow)98.6Omni-sourced Webly-supervised Learning for Video Recognition
PERF-Net (multi-distilled S3D)98.6PERF-Net: Pose Empowered RGB-Flow Net-
ZeroI2V ViT-L/1498.6ZeroI2V: Zero-Cost Adaptation of Pre-trained Transformers from Image to Video
LGD-3D Two-stream98.2Learning Spatio-Temporal Representation with Local and Global Diffusion-
Text4Vis98.2Revisiting Classifier: Transferring Vision-Language Models for Video Recognition
Two-Stream I3D (Imagenet+Kinetics pre-training)98.0Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
Two-Stream I3D (Kinetics pre-training)97.8Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
MARS+RGB+Flow (64 frames, Kinetics pretrained)97.8MARS: Motion-Augmented RGB Stream for Action Recognition-
HATNet (32 frames)97.8Large Scale Holistic Video Understanding
BubbleNET97.62Bubblenet: A Disperse Recurrent Structure To Recognize Activities-
BQN97.6Busy-Quiet Video Disentangling for Video Classification
D3D + D3D97.6D3D: Distilled 3D Networks for Video Action Recognition
CCS + TSN (ImageNet+Kinetics pretrained)97.4Cooperative Cross-Stream Network for Discriminative Action Representation-
R[2+1]D-TwoStream (Kinetics pretrained)97.3A Closer Look at Spatiotemporal Convolutions for Action Recognition
0 of 90 row(s) selected.
Action Recognition In Videos On Ucf101 | SOTA | HyperAI超神经