How to compare the predicted boundary-map crispness?
xavysp opened this issue · 0 comments
xavysp commented
Dear @Wangyupei,
I really enjoy reading your paper. And looking at the CED prediction and the others from HED and RCF, your results are crisper (according to my eyes). However, I still do not follow (Pages 7 and 9 in the journal paper) about the quantitative comparison for the crispness. Please, could you give me more details?
Do you think that if I evaluate the prediction without NMS and setting d= d_0/4, could I obtain a fair comparison for the crispness evaluation?
Xavier