Welcome to openSVAI’s documentation!¶
openSVAI is a project aimed at:
- open-sourcing utility code and bash scripts to reduce repetitive implementations and environment set-ups from SVAI members;
- providing various end-to-end module baselines for common tasks, e.g., human candidates detection, keypoints regression and pose tracking;
- standardizing end-to-end module outputs so that different methods may co-exist in a single module;
- implementing a centralized program that calls different modules based on a configuration file, the modules of which can be from different deep learning frameworks.
\[1 + 1 > 2\]
Documentation¶
- Installation of Components
- Utilities
- Standardization
- Format Examples
- Output Detections To SVAI Standard Json
- Read Detections From SVAI Standard Json
- Output Keypoints To SVAI Standard Json
- Read Both Detections and Keypoints From SVAI Standard Json
- Output Pose Tracking Results To SVAI Standard Json
- Visualizer
- Conversion Between Formats
- Other Scripts: Batch Testing
- AR applications
- ECCV 2018 papers