search

GuoLanqing / ShadowFormer

ShadowFormer (AAAI2023), Pytorch implementation
MIT License
129 stars 17 forks source link

Testing with custom data #5

Closed Alive59 closed 1 year ago

Alive59 commented 1 year ago

Is it possible for testing using data without ground truth (test_B, test_C folders) ?

GuoLanqing commented 1 year ago

You can use any existing shadow detector, e.g., DHAN (AAAI2020), to predict the shadow mask. Then input the testing image and corresponding predicted shadow mask into the network for testing. The test_C folder is GT, which is not compulsory if you do not want to calculate the metrics.

Alive59 commented 1 year ago

You can use any existing shadow detector, e.g., DHAN (AAAI2020), to predict the shadow mask. Then input the testing image and corresponding predicted shadow mask into the network for testing. The test_C folder is GT, which is not compulsory if you do not want to calculate the metrics.

Okay I see. Thanks for the quick reply!