研究目的
To perform recognition and 3D localization of dynamic objects on images from a stereo camera for traffic scene understanding in urban and road environments.
研究成果
The proposed system demonstrates satisfactory results under different setups, achieving robust and accurate detection of objects, extraction of additional features for high-level inference, and real-time performance. The method provides valuable information for a higher-level understanding of the traffic scene.
研究不足
The accuracy in depth estimation increases with the distance to the camera, and the method assumes a flat ground which may not hold in all driving environments.
