About this document
3D Vehicle Tracking with Self-Calibration by fellhuber.maximilian is a document available to read on EtoBox.
The document presents a method for tracking multiple vehicles using a 3D deformable model and camera self-calibration to address occlusion issues in traffic surveillance. It employs a constrained multiple-kernel tracking approach, utilizing both shape and color information, and integrates object detection techniques from SSD and YOLO9000 to enhance accuracy. The proposed system demonstrates improved tracking performance in complex urban environments, leveraging evolutionary algorithms for camera parameter o
- Author
- fellhuber.maximilian
- Language
- EN