search

zhang-tao-whu / e2ec

E2EC: An End-to-End Contour-based Method for High-Quality High-Speed Instance Segmentation
Other
214 stars 45 forks source link

training problem #6

Closed xubangwu closed 2 years ago

xubangwu commented 2 years ago

Hello, when we train on the KITTI dataset, the input size is (384,896) or (512,896)?

zhang-tao-whu commented 2 years ago

The input size for training on KITTI is (384, 896). The size of the network input image is cropped to the (input_h, input_w) in the config file. In the config file of KITTI, data.input_w, input_h = (896, 384).