Open killwy opened 2 years ago
Hello, the released model is designed for cross-domain and joint generalization. The model employed for the finetuning performance on KITTI2015 is a little different from the released code. You can see section 4.5 Results on KITTI Benchmark of our paper for more details.
Can you provide the model trained by kitti2015? I could not achieve the results in the paper by reproduction