Activity-Recognition and Video-Understanding

This repository contains a compilation of code implementations for numerous works in Activity Recognition.

Below is a general purpose template for Activity Recognition:

Visual Attributions:

1. Video attributions Code

General code bases:

1. SlowFast by Facebook Research Link

2. MMAction Link
3. MMAction2 Link
4. ClassyVision Link
5. Facebook Video Modelling Zoo Link
6. PyVideoResearch Link
7. GluonCV Link
8. M-PACT Link
9. PyTorch Video Recognition Link

Multi-stream Methods:

1. Convolutional Two-Stream Network Fusion for Video Action Recognition Code | Paper
2. Temporal Segment Networks: Towards Good Practices for Deep Action Recognition Code | Paper
3. ActionVLAD: Learning spatio-temporal aggregation for action classification Code | Paper
4. Hidden Two-Stream Convolutional Networks for Action Recognition Code | Paper
5. Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset Code | Code | Code | Paper

Single-Stream Methods:

LSTM Methods:

1. Long-term Recurrent Convolutional Networks for Visual Recognition and Description Code | Paper
2. What Would You Expect? Anticipating Egocentric Actions With Rolling-Unrolling LSTMs and Modality Attention Code | Paper
3. TS-LSTM and Temporal-Inception: Exploiting Spatiotemporal Dynamics for Activity Recognition Code | Paper

CNN Methods:

1. Temporal 3D ConvNets: New Architecture and Transfer Learning for Video Classification Code | Paper
2. TSM: Temporal Shift Module for Efficient Video Understanding Code | Paper
3. Temporal Relational Reasoning in Videos Code | Paper
4. Non-Local Neural Networks Code | Paper
5. Video Classification with Channel-Separated Convolutional Networks Unofficial Code | Paper
6. Gate-Shift Networks for Video Action Recognition Code | Paper
7. V4D: 4D Convolutional Neural Networks for Video-level Representation Learning Code | Paper
8. Temporal Interlacing Network Code | Paper

Attention-based Methods:

1. Action Recognition using Visual Attention Code | Code | Paper
2. Attention Clusters: Purely Attention Based Local Feature Integration for Video Classification Unofficial Code | Paper
3. Video Modeling with Correlation Networks Code | Paper
4. Attentional Pooling for Action Recognition Code | Paper

Miscellenous:

1. Temporal Convolutional Networks: A Unified Approach to Action Segmentation and Detection Code | Paper
2. Long-term Temporal Convolutions Code | Paper
3. Dynamic Image Networks for Action Recognition Code | Paper
4. Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks Code | Paper
5. Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet? Code | Paper
6. Long-Term Feature Banks for Detailed Video Understanding Code | Paper
7. Learning Correspondence from the Cycle-consistency of Time Code | Paper
8. Why Can't I Dance in the Mall? Learning to Mitigate Scene Bias in Action Recognition Code | Paper
9. Learning Actor Relation Graphs for Group Activity Recognition Code | Paper
10. Asynchronous Temporal Fields for Action Recognition Code | Paper
11. TEA: Temporal Excitation and Aggregation for Action Recognition Code | Paper
12. MotionSqueeze: Neural Motion Feature Learning for Video Understanding Code | Paper
13. ECO: Efficient Convolutional Network for Online Video Understanding Code | Paper
14. End-to-End Learning of Motion Representation for Video Understanding Code | Paper
15. AR-Net: Adaptive Frame Resolution for Efficient Action Recognition Code | Paper
16. Learn to cycle: Time-consistent feature discovery for action recognition Code | Paper
17. VideoGraph: Recognizing Minutes-Long Human Activities in Videos Code | Paper
18. Timeception for Complex Action Recognition Code | Paper
19. An Evaluation of Action Recognition Models on EPIC-Kitchens Code | Paper
20. STEP: Spatio-Temporal Progressive Learning for Video Action Detection Code | Paper
21. Appearance-and-Relation Networks for Video Classification Code | Paper
22. End-to-end Video-level Representation Learning for Action Recognition Code | Paper
23. Action Recognition with Trajectory-Pooled Deep-Convolutional Descriptors Code | Paper
24. Real-time Action Recognition with Enhanced Motion Vector CNNs Code | Paper
25. Temporal-Relational CrossTransformers for Few-Shot Action Recognition Code | Paper