Action Classification
Action Classification is an important sub-task in the field of computer vision, focusing on recognizing and categorizing human actions in videos. This task aims to accurately classify different types of actions into predefined categories by analyzing dynamic features in video sequences, thereby achieving automatic understanding of human activities. Its application value is extensive, including but not limited to intelligent surveillance, human-computer interaction, sports analysis, and other fields, which can significantly enhance the intelligence level of systems and user experience.
Kinetics-400
MTV-H (WTS 60M)
Kinetics-600
MViT-B-24, 32x3
Charades
TokenLearner
Kinetics-700
MoViNet-A6
Toyota Smarthome dataset
π-ViT
AViD
TokenLearner
Moments in Time
ActivityNet-1.2
W-TALC
Kinetics-700-2020
ALIP-ViT B/32 LAION30M
THUMOS’14
3C-Net
WiGesture
Kinetics-Sounds
MIT
InternVideo2-6B
TTStroke-21 ME22
RGB and PRGB
ActivityNet
UniFormerV2-L
BABEL
2s-AGCN
CelebV-HQ
Diving-48
DualPath w/ ViT-B/16
HMDB51
Jester test
MiniKinetics
MARS+RGB+Flow (16 frames)
Something-Something V2
AdaMAE
THUMOS'14
3C-Net
TTStroke-21 ME21
UCF101
Ours
YouCook2
VideoBERT (cross modal)