JXingZhao / EGNet

EGNet:Edge Guidance Network for Salient Object Detection (ICCV 2019)
311 stars 62 forks source link

Could you please tell me how to get the edge label to supervise the model while training? #2

Open pengqianli opened 4 years ago

hongyann commented 4 years ago

作者你好,请问您使用的canny边缘检测算法是将saliency map 作为输入吗?我查到的canny算法基本都是彩色图片作为输入,所以如果可以的话,方便提供一份您使用的canny边缘提取的算法代码吗? 我的邮箱hongyann.cai@gmail.com

期待您的回复,万分感谢!

JXingZhao commented 4 years ago

I am sorry that I misunderstand the question. We use the canny algorithm to get the edge map from the segmentation results of other methods. During training, we use the simple gradient to get the ground-truth edge label. The core code is followed: [gy, gx] = gradient(gt); edge = gy.gy + gx.gx; edge(edge~=0)=1; It is implemented by matlab.

hongyann commented 4 years ago

thank you!

JXingZhao notifications@github.com 于2019年9月1日周日 下午8:16写道:

I am sorry that I misunderstand the question. We use the canny algorithm to get the edge map from the segmentation results of other methods. During training, we use the simple gradient to get the ground-truth edge label. The core code is followed: [gy, gx] = gradient(gt); edge = gy.gy + gx.gx; edge(edge~=0)=1; It is implemented by matlab.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JXingZhao/EGNet/issues/2?email_source=notifications&email_token=AMWMHREDVQIOMV7UAQXJUMLQHOXBTA5CNFSM4ISMVNOKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5UBD7A#issuecomment-526914044, or mute the thread https://github.com/notifications/unsubscribe-auth/AMWMHRAYTWVCQ7WT4ST3FCDQHOXBTANCNFSM4ISMVNOA .

jiwei0921 commented 4 years ago

作者你好,请问您使用的canny边缘检测算法是将saliency map 作为输入吗?我查到的canny算法基本都是彩色图片作为输入,所以如果可以的话,方便提供一份您使用的canny边缘提取的算法代码吗? 我的邮箱hongyann.cai@gmail.com

期待您的回复,万分感谢!

You also can use the mask to produce edge maps. The link is https://blog.csdn.net/qq_19329785/article/details/100077984

JXingZhao commented 4 years ago

作者你好,请问您使用的canny边缘检测算法是将saliency map 作为输入吗?我查到的canny算法基本都是彩色图片作为输入,所以如果可以的话,方便提供一份您使用的canny边缘提取的算法代码吗? 我的邮箱hongyann.cai@gmail.com 期待您的回复,万分感谢!

You also can use the mask to produce edge maps. The link is https://blog.csdn.net/qq_19329785/article/details/100077984

Yes, you are right. This step is very simple.

hongyann commented 4 years ago

thankyou