Skip to content

Latest commit

 

History

History

action_recognition

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
readme.md
readme.md
 
 

readme.md

List for public implementation of various algorithms

1) Pubilc Datasets and Challenges

2) Pioneers and Experts

3) Blogs and Videos

4) Papers and Sources Codes

▶ Datasets Papers

  • KTH(ICPR2004) Recognizing human actions: a local SVM approach [paper link]
  • Weizmann(ICCV2005) Actions as space-time shapes [paper link]
  • UCF101(arxiv2012) UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild [arxiv link]
  • Kinetics(arxiv2017) The Kinetics Human Action Video Dataset [arxiv link]
  • EPIC-Kitchens(ECCV2018) Scaling Egocentric Vision: The EPIC-KITCHENS Dataset [project link]
  • HACS(arxiv2019) HACS: Human Action Clips and Segments Dataset for Recognition and Temporal Localization [arxiv link]
  • Moments-in-Time(TPAMI2019) Moments in Time Dataset: one million videos for event understanding [project link]
  • FineGym(CVPR2020) FineGym: A Hierarchical Video Dataset for Fine-grained Action Understanding [project link]

▶ Technique Papers

1) 基于人工特征(Manual-Features)

2) 基于时空双流神经网络(Two-Stream)

  • Two-Stream(NIPS2014) Two-Stream Convolutional Networks for Action Recognition in Videos [arxiv link]

  • two-stream+LSTM(CVPR2015) Long-term Recurrent Convolutional Networks for Visual Recognition and Description [arxiv link][project link][Codes|offical]

  • two-stream+LSTM(CVPR2015) Beyond short snippets: Deep networks for video classification [paper link]

  • two-stream fusion(CVPR2016) Convolutional Two-Stream Network Fusion for Video Action Recognition [arxiv link][Codes|offical Matlab MatConvNet]

  • TSN(ECCV2016) Temporal Segment Networks: Towards Good Practices for Deep Action Recognition [arxiv link][project link][Codes|PyTorch(offical)]

  • Co-occurrence+LSTM(+pose)(AAAI2016) Co-occurrence Feature Learning for Skeleton based Action Recognition using Regularized Deep LSTM Networks [arxiv link]

  • RNN-based(+pose)(ECCV2016) Online Human Action Detection using Joint Classification-Regression Recurrent Neural Networks [arxiv link]

  • TSN-based improved 1(CVPR2017) Deep Local Video Feature for Action Recognition [arxiv link]

  • ST+Attention+LSTM(+pose)(AAAI2017) An End-to-End Spatio-Temporal Attention Model for Human Action Recognition from Skeleton Data [arxiv link]

  • TRN(TSN-based improved 2)(ECCV2018) Temporal Relational Reasoning in Videos [arxiv link]

  • ST-GCN(+openpose)(AAAI2018) Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition [arxiv link]

  • 密集扩张网络(TIP2019) Dense Dilated Network for Video Action Recognition [paper link]

3) 基于三维卷积的神经网络(3D-ConvNet)

4) 基于长短记忆网络(LSTM)

5) 基于对抗神经网络(GAN)

  • GAN-based(IJCAI2018) Exploiting Images for Video Recognition with Hierarchical Generative Adversarial Networks [arxiv link]