Sourcify
Effortlessly find and manage open-source dependencies for your projects.
Development
Freemium
From $99/mo
Smart Contract Verification
Dependency Management
Security Auditing
TimeSformer
4.6
Free Highly accurate and efficient for video understanding tasks, with a strong focus on state-of-the-art performance.
Space-Time Attention for Video Understanding.
0 views
0 saves
|
Refined 2/22/2026
About TimeSformer
TimeSformer is a PyTorch-based video classification framework developed by Facebook Research. It leverages space-time attention mechanisms to achieve state-of-the-art results in video action recognition. The model processes video by dividing it into space and time components, applying self-attention to each, and then combining the results. This architecture enables efficient processing of long video sequences while maintaining high accuracy. Key use cases include action recognition, video content analysis, and video search. It is designed to provide flexibility in terms of different self-attention schemes. The framework includes pre-trained models on Kinetics-400, Kinetics-600, Something-Something-V2, and HowTo100M datasets.
Core Capabilities
TimeSformer is a PyTorch-based video classification framework developed by Facebook Research.
Main Tasks
Action Recognition
Explore all tools that specialize in action recognition. This domain focus ensures TimeSformer delivers optimized results for this specific requirement.
