Project / 03 · 2025—26
TSMTemporal vision / HRI
Real-time Hand Gesture Recognition for Human–Robot Collaboration
A Temporal Shift Module pipeline for real-time gesture recognition, with data preparation, training configuration, checkpoint loading, and webcam inference.
Research question
How can a lightweight temporal model turn a sequence of human gestures into an efficient interaction signal for a collaborative robot?
This page treats the project as a system and a research record, not as a list of technologies.
System
Method & contribution
Implemented the TSM training and real-time deployment path for a custom five-class gesture task, including frame preparation, configuration, checkpoint loading, and webcam demonstration.
- 实现 TSM 训练与实时部署路径
- 负责帧准备、配置、checkpoint 加载和 webcam 演示
- 将公开仓库范围限定在自有视觉模块
- Adapt a temporal-shift implementation to a custom dataset
- Keep the public release focused on the owned vision contribution
- Make training and deployment steps reproducible
- Separate the larger team pipeline from this module
Evidence
- Training and evaluation path documented for reproducible runs
- Real-time webcam entry point preserved in the public module
- Public source repository
- English and Chinese documentation
- Training/evaluation entry points
- Real-time webcam demo path
Robot grasping, object detection, segmentation, pose estimation, and robot control belong to the larger team project and are not claimed by this repository.