search

zzangjinsun / NLSPN_ECCV20

Park et al., Non-Local Spatial Propagation Network for Depth Completion, ECCV, 2020
MIT License
321 stars 55 forks source link

question on kitti #19

Closed Xuanmeng-Zhang closed 3 years ago

Xuanmeng-Zhang commented 3 years ago

Hi, Park. I run the code in kitti dataset with batch size 12 for 20 epochs on 4 gpus. But the rmse on validation dataset is 781mm. Maybe something is wrong during my training? Could you share me the files saved in the save_dir? Thank you.

zzangjinsun commented 3 years ago

I remember that the best result I got was with a batch size 24 on 6 GPUs. If you want to use the best model, please refer to the released models: https://github.com/zzangjinsun/NLSPN_ECCV20#pre-trained-models-and-results