search

xw-hu / DSC

CVPR 2018 (Oral) & TPAMI 2019
141 stars 35 forks source link

How to convert the predicted map to a binary image for evaluation? #1

Closed ShuhanChen closed 6 years ago

ShuhanChen commented 6 years ago

I found the predicted shadow map is an intensity image, how to convert it to a binary image for evaluation? Just set a global threshold, e.g., 0?

xw-hu commented 6 years ago

We apply CRF as the post-processing in testing. The predicted map after CRF is a binary map. You can find the link of the code in README/Test.

ShuhanChen commented 6 years ago

I downloaded your result maps, and found that there are still some pixel values between 0~255, e.g. 624.jpg. So I can't get the same accuracy and BER value as listed in your paper. Can you share the evaluation code? In fig.5 of your paper, the last three columns also seem like intensity maps. Thanks for your quick reply.

xw-hu commented 6 years ago

Maybe some images I forgot to apply the CRF. Anyway, the results should be similar. Please give me your email and I send you the evaluation code.

ShuhanChen commented 6 years ago

c.shuhan@gmail.com thanks very much

quanlzheng commented 6 years ago

Dear Xiaowei,

I am interested in your work. I test your results but found the results have some difference in your paper, could you share the evaluation code with me for better evaluation?

My email is xiaolong921001@gmail.com

Thanks very much