search

JizhiziLi / AIM

[IJCAI'21] Deep Automatic Natural Image Matting
MIT License
386 stars 34 forks source link

Wrong SAD result #19

Open gaka2012 opened 2 years ago

gaka2012 commented 2 years ago

I tested the aim-net500 dataset with the model you provided and found that the SAD was only 47. Can you check that the model you uploaded in GitHub is correct? In addition, did you do other processing on the DIM dataset? In your dim_hatt_am2k_type.json file, the key of the DIM dataset is number 1,2,3,4....，but the names of DIM datasets are different. So how did you rename them. I use the training_fg_names.txt file to rename them, and retrained the model according to the steps you provided on GitHub, and the SAD is only 55.

bah111587 commented 2 years ago

I tested the aim-net500 dataset with the model you provided and found that the SAD was only 47. Can you check that the model you uploaded in GitHub is correct? In addition, did you do other processing on the DIM dataset? In your dim_hatt_am2k_type.json file, the key of the DIM dataset is number 1,2,3,4....，but the names of DIM datasets are different. So how did you rename them. I use the training_fg_names.txt file to rename them, and retrained the model according to the steps you provided on GitHub, and the SAD is only 55.

hello,I found that my dim_hatt_am2k_type.json file downloaded from the github have the same problem.could you pleased to provide me for your renamed json file? My email is 1147640900@qq.com. Thank you very much!