Action Recognition In Videos On Ucf101

Metriken

3-fold Accuracy

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	3-fold Accuracy	Paper Title	Repository
Slow Fusion + Finetune top 3 layers	65.4	Large-Scale Video Classification with Convolutional Neural Networks
HalluciNet (ResNet-50)	79.83	HalluciNet-ing Spatiotemporal Representations Using a 2D-CNN
I3D RGB + DMC-Net (I3D)	96.5	DMC-Net: Generating Discriminative Motion Cues for Fast Compressed Video Action Recognition	-
LGD-3D RGB	97	Learning Spatio-Temporal Representation with Local and Global Diffusion	-
I3D-LSTM	95.1	I3D-LSTM: A New Model for Human Action Recognition	-
VIMPAC	92.7	VIMPAC: Video Pre-Training via Masked Token Prediction and Contrastive Learning
Two-Stream I3D (Kinetics pre-training)	97.8	Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
MV-CNN	86.4	Real-time Action Recognition with Enhanced Motion Vector CNNs
ST-ResNet + IDT	94.6	Spatiotemporal Residual Networks for Video Action Recognition
R[2+1]D-Flow (Sports-1M pretrained)	93.3	A Closer Look at Spatiotemporal Convolutions for Action Recognition
SVT	93.7	Self-supervised Video Transformer
ResNet50	-	Adaptive frame selection in two dimensional convolutional neural network action recognition
MLGCN	63.27	MLGCN: Multi-Laplacian Graph Convolutional Networks for Human Action Recognition	-
CMA iter1-S	96.5	Two-Stream Video Classification with Cross-Modality Attention	-
R[2+1]D-RGB (Sports-1M pretrained)	93.6	A Closer Look at Spatiotemporal Convolutions for Action Recognition
Flow-I3D (Imagenet+Kinetics pre-training)	96.7	Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
Hidden Two-Stream	97.1	Hidden Two-Stream Convolutional Networks for Action Recognition
OmniVec2	99.6	OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning	-
MF-Net, RGB only (ImageNet+Kinetics pretrained)	96.0	Multi-Fiber Networks for Video Recognition	-
OmniVec	99.6	OmniVec: Learning robust representations with cross modal sharing	-

0 of 90 row(s) selected.