zengxianyu / lps

code for the paper

lps

code for the paper ``learning to promote saliency detectors"

Environment: python 2.7; pytorch '0.5.0a0+54db14e'; two GTX 1080Ti GPU;

usage

modify the path to prior maps, images, groud truth and then run test.py

pre-trained model

download pre-trained model

performance and results

This version gives slightly different results from the paper.

Prior maps are the results of other methods which are provided by the authors or obtained by runing their code. For example, results of SRM can be downloaded here. Results of applying our method on SRM can be downloaded from 百度网盘 or google drive

F-measure

v	ECSSD	HKU-IS	PASCALS	DUTS-test	THUR	OMRON
SRM	.8924	.8739	.7961	.7591	.7079	.7223
+Ours	.9102	.9032	.8054	.7999	.7299	.7338

MAE

v	ECSSD	HKU-IS	PASCALS	DUTS-test	THUR	OMRON
SRM	.0542	.0459	.0852	.0633	.0769	.0767
+Ours	.0416	.0330	.0729	.0536	.0735	.0696

I'll add the results on other methods later.