Vision-based risk quantification on E-bikes for urban cycling
Vision-based risk quantification on E-bikes for urban cycling
In my master’s thesis project at the Computer Vision Lab, ETH Zurich, I focused on:
- Isovist Generation:
- Create isovists using edge detection.
- Incorporate big data analysis for accuracy.
- Implement dynamic navigation with real-time GPS.
- Human Pose Estimation:
- Optimize human pose estimation for embedded systems.
- Use Nvidia Jetson and TensorRT for real-time performance.
- Risk Maps Generation:
- Estimate traffic accident probabilities based on detected pedestrain density.
- Generate risk maps for urban safety based on estimated traffic accidents.
This project leveraged advanced computer vision, embedded systems, and data analysis to enhance urban navigation and safety.