CV_project
Final project in Computer Vision course of ZJU. Based on the aerial data of zijingang Campus buildings, three tasks were completed: visual positioning (feature matching and camera pose estimation using deep learning method), dense reconstruction (3d dense reconstruction and depth estimation), and semantic segmentation (2d and 3D semantic segmentation using ensemble of nine existing models)
- visual positioning ： loc
- dense reconstruction：dpt
- semantic segmentation：seg