EGNet icon indicating copy to clipboard operation
EGNet copied to clipboard

For testing:which path should I change?

Open wotaiaixx opened this issue 5 years ago • 7 comments

In the dataset.py ,have much path,which which path should I change for the one picture? I can't find the test.lst file.

thanks for you read.

wotaiaixx avatar Aug 30 '19 05:08 wotaiaixx

Or you could view it as how to generate this .lst

wotaiaixx avatar Aug 30 '19 05:08 wotaiaixx

For on image, what you should do is that you add the several lines in the line 86 of dataset.py.

For example, if you picture are called 1.jpg and in /data/test/img/. You should generate the test.lst first by ls command. Then you can add these code: else if sal_mode == 'abc': self.image_root = '/data/test/img/' self.image_source = '/data/test/test.lst' self.test_fold = '/data/test/' Then you run the python3 run.py --mode test --sal_mode abc You will get the result in /data/test/

发件人:wotaiaixx [email protected] 发送日期:2019-08-30 13:53:20 收件人:JXingZhao/EGNet [email protected] 抄送人:Subscribed [email protected] 主题:Re: [JXingZhao/EGNet] For testing:which path should I change? (#1) Or you could view it as how to generate this .lst — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

JXingZhao avatar Aug 30 '19 06:08 JXingZhao

Hello, first, I need to emphasize that I just doing I a picture test. I have already slove all the procedural problem,but met CUDA out of memory,CUDA=4G then, I close the CUDA=Flase ,use CPU run this code,was killed. what should i do for this problem ?

qqwinn avatar Aug 30 '19 12:08 qqwinn

Hi, it maybe because of the limitation of cuda memory. You can try to run it by the GPU with larger memory.

发件人:qqwinn [email protected] 发送日期:2019-08-30 20:13:15 收件人:JXingZhao/EGNet [email protected] 抄送人:JXingZhao [email protected],Comment [email protected] 主题:Re: [JXingZhao/EGNet] For testing:which path should I change? (#1) Hello, first, I need to emphasize that I just doing I a picture test. I have already slove all the procedural problem,but met CUDA out of memory,CUDA=4G then, I close the CUDA=Flase ,use CPU run this code,was killed. what should i do for this problem ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

JXingZhao avatar Aug 30 '19 12:08 JXingZhao

If i use epoch_vgg.pth weight replace epoch_resnet.pth, where should i change ?

qqwinn avatar Aug 30 '19 12:08 qqwinn

Just change the 25 line in solver.py

JXingZhao avatar Aug 31 '19 01:08 JXingZhao

For on image, what you should do is that you add the several lines in the line 86 of dataset.py. For example, if you picture are called 1.jpg and in /data/test/img/. You should generate the test.lst first by ls command. Then you can add these code: else if sal_mode == 'abc': self.image_root = '/data/test/img/' self.image_source = '/data/test/test.lst' self.test_fold = '/data/test/' Then you run the python3 run.py --mode test --sal_mode abc You will get the result in /data/test/ 发件人:wotaiaixx [email protected] 发送日期:2019-08-30 13:53:20 收件人:JXingZhao/EGNet [email protected] 抄送人:Subscribed [email protected] 主题:Re: [JXingZhao/EGNet] For testing:which path should I change? (#1) Or you could view it as how to generate this .lst — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

I get nothing.

cs19469 avatar Dec 29 '21 08:12 cs19469