search

cvlab-stonybrook / DM-Count

Code for NeurIPS 2020 paper: Distribution Matching for Crowd Counting.
MIT License
218 stars 52 forks source link

About the min_size, max_size in the preprocess #28

Open zhou-rui1 opened 2 years ago

zhou-rui1 commented 2 years ago

Hi, thanks for this meaningful work. I am wondering how will the min_size, max_size effects the final results or the training process?

https://github.com/cvlab-stonybrook/DM-Count/blob/cc5f2132e0d1328909f31b6d665b8e0b15c30467/preprocess/preprocess_dataset_nwpu.py#L99

zhou-rui1 commented 2 years ago

New experients shows that, when I train it with larger input size(first assign larger min_size), Wass Distance appear to be nan...thus the overall mae be higher...

could you give me some guidance on this problem?

with best regards, rui