search

taozh2017 / SPNet

Specificity-preserving RGB-D Saliency Detection
47 stars 8 forks source link

Performance on DUT #4

Closed Angknpng closed 2 years ago

Angknpng commented 2 years ago

Have you evaluted the model on DUT-RGBD? I have used the code test_performance_maps.py and the model 'SPNet_best.pth' you provided to predicet the salient maps. However, the results is different from you gived in the paper, including the NJUD. Is there anything wrong? I hope for your reply! Thanks!

taozh2017 commented 2 years ago

Which evaluation toolbox do you use? There is a little difference between Python evaluation and Matlab version.

Angknpng commented 2 years ago

I used the Matlab valuation toolbox. In general, there is little difference between the two Toolbox, but there is a big difference between the results I tested here. I suspect that the model you provided is not optimal, or there is something wrong there. Could you provide the predicted results on DUT?

------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年3月30日(星期三) 上午10:05 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4)

Which evaluation toolbox do you use? There is a little difference between Python evaluation and Matlab version.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

Angknpng commented 2 years ago

This is the test results I got on Matlab Evaluation Toolbox. The NLPR result is very close to what you gave, but the other two are quite different from what you gave. But I tested the predictions using the code and model you provided on github. For a fair comparison, I hope you can provide the results of DUT

------------------ 原始邮件 ------------------ 发件人: "taozh2017/SPNet" @.>; 发送时间: 2022年3月30日(星期三) 上午10:05 @.>; @.**@.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4)

Which evaluation toolbox do you use? There is a little difference between Python evaluation and Matlab version.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

taozh2017 commented 2 years ago

OK. When I am free and I test it on DUT. I can also show the results using the Matlab evaluation toolbox, the difference is very small.

Angknpng commented 2 years ago

Thank you for your reply and I really appreciate the SPNet you designed. I look forward to your results presentation.

------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年3月30日(星期三) 中午1:22 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4)

OK. When I am free and I test it on DUT. I can also show the results using the Matlab evaluation toolbox, the difference is very small.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

taozh2017 commented 2 years ago

Thank you for your reply and I really appreciate the SPNet you designed. I look forward to your results presentation. … ------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年3月30日(星期三) 中午1:22 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4) OK. When I am free and I test it on DUT. I can also show the results using the Matlab evaluation toolbox, the difference is very small. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

Can you provide your email? and then I send you the predicted maps of DUT-RGBD. Thanks.

Angknpng commented 2 years ago

Thank you! my email is @.***

------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年4月6日(星期三) 晚上8:39 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4)

Thank you for your reply and I really appreciate the SPNet you designed. I look forward to your results presentation. … ------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年3月30日(星期三) 中午1:22 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4) OK. When I am free and I test it on DUT. I can also show the results using the Matlab evaluation toolbox, the difference is very small. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

Can you provide your email? and then I send you the predicted maps of DUT-RGBD. Thanks.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

taozh2017 commented 2 years ago

Your email is blocked. You go to my website and find my email.

Angknpng commented 2 years ago

OK. I will send a e-mail to you.

------------------ 原始邮件 ------------------ 发件人: "Tao @.>; 发送时间: 2022年4月6日(星期三) 晚上9:20 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [taozh2017/SPNet] Performance on DUT (Issue #4)

Your email is blocked. You go to my website and find my email.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>