Welcome to MMAction2’s documentation!

You can switch between Chinese and English documents in the lower-left corner of the layout.

• Installation
  • Requirements
  • Install MMAction2
  • Install with CPU only
  • Another option: Docker Image
  • A from-scratch setup script
  • Developing with multiple MMAction2 versions
  • Verification
• Getting Started
  • Datasets
  • Inference with Pre-Trained Models
  • Build a Model
  • Train a Model
  • Tutorials
• Demo
  • Demo link
  • Video demo
  • SpatioTemporal Action Detection Video Demo
  • Video GradCAM Demo
  • Webcam demo
  • Long video demo
• Benchmark
  • Settings
  • Main Results
  • Details of Comparison

Datasets

• Overview
  • Supported Datasets
• Data Preparation
  • Notes on Video Data Format
  • Getting Data
• Supported Datasets
  • ActivityNet
  • AVA
  • GYM
  • HMDB51
  • HVU
  • Jester
  • JHMDB
  • Kinetics-[400/600/700]
  • Moments in Time
  • Multi-Moments in Time
  • OmniSource
  • Something-Something V1
  • Something-Something V2
  • THUMOS’14
  • UCF-101
  • UCF101-24

Model Zoo

• Overview
  • Spatio Temporal Action Detection Models
  • Action Localization Models
  • Action Recognition Models
• Action Recognition Models
  • C3D
  • CSN
  • I3D
  • Omni-sourced Webly-supervised Learning for Video Recognition
  • R2plus1D
  • SlowFast
  • SlowOnly
  • TANet
  • TIN
  • TPN
  • TRN
  • TSM
  • TSN
  • X3D
  • ResNet for Audio
• Action Localization Models
  • BMN
  • BSN
  • SSN
• Spatio Temporal Action Detection Models
  • AVA
  • LFB

Tutorials

• Tutorial 1: Learn about Configs
  • Modify config through script arguments
  • Config File Structure
  • Config File Naming Convention
  • FAQ
• Tutorial 2: Finetuning Models
  • Outline
  • Modify Head
  • Modify Dataset
  • Modify Training Schedule
  • Use Pre-Trained Model
• Tutorial 3: Adding New Dataset
  • Customize Datasets by Reorganizing Data
  • Customize Dataset by Mixing Dataset
• Tutorial 4: Customize Data Pipelines
  • Design of Data Pipelines
  • Extend and Use Custom Pipelines
• Tutorial 5: Adding New Modules
  • Customize Optimizer
  • Customize Optimizer Constructor
  • Develop New Components
  • Add new learning rate scheduler (updater)
• Tutorial 6: Exporting a model to ONNX
  • Supported Models
  • Usage
• Tutorial 7: Customize Runtime Settings
  • Customize Optimization Methods
  • Customize Training Schedules
  • Customize Workflow
  • Customize Hooks

Useful Tools and Scripts

• Useful Tools Link
• Log Analysis
• Model Complexity
• Model Conversion
  • MMAction2 model to ONNX (experimental)
  • Prepare a model for publishing
• Miscellaneous
  • Evaluating a metric
  • Print the entire config

Notes

• Changelog
  • Master
  • 0.13.0 (31/03/2021)
  • 0.12.0 (28/02/2021)
  • 0.11.0 (31/01/2021)
  • 0.10.0 (31/12/2020)
  • 0.9.0 (30/11/2020)
  • v0.8.0 (31/10/2020)
  • v0.7.0 (30/9/2020)
  • v0.6.0 (2/9/2020)
  • v0.5.0 (9/7/2020)
• FAQ
  • Installation
  • Data
  • Training
  • Testing
  • Deploying

API Reference

• API Reference
  • mmaction.apis
  • mmaction.core
    • optimizer
    • evaluation
    • lr
  • mmaction.localization
    • localization
  • mmaction.models
    • models
    • recognizers
    • localizers
    • common
    • backbones
    • heads
    • necks
    • losses
  • mmaction.datasets
    • datasets
    • pipelines
    • samplers
  • mmaction.utils
  • mmaction.localization

Switch Language

• English
• 简体中文

Indices and tables

• Index

• Search Page