How to train my dataset using FCN

[YangFangShu]

Hi,
Thanks for every one ,I am a newer to FCN and I want to use FCN to train my dataset. The dataset contains original images which are 12000*(800801)(they are gray figure) and labeled
images which are 12000*(80*80).Because the original images and labeled images are different images, but they have non-linear relationship, and in labeled images, I want to identify 2 classes that the interest(labeled as 1)and background(labeled as 0),so I want to use FCN to learn the non-linear algorithm and identify two parts in labeled images. I have some trouble in how can I do to solve this problem, I think I should convert my data to "imdb.mat" to feed fcntrain firstly, but I didn't achieve it .Then I think I should run fcnInitializeModel to initialize the net and modify some parameters in the code.What I think is right?I hope someone can help me ,and I am grateful if you can share your code to me

Best Regards
Email:yfs2016@hit.edu.cn

[steeveen]

老铁哈工的？

[zhenghaojie666]

@YangFangShu I meet same problem, have you slove this error.
thank you.

[JoshuaLPF]

Thanks for every one ,I am a newer to FCN and I want to use FCN to train my dataset. The dataset contains original images which are 12000*(800_80_1)(they are gray figure) and labeled
images which are 12000*(80*80).Because the original images and labeled images are different images, but they have non-linear relationship, and in labeled images, I want to identify 2 classes that the interest(labeled as 1)and background(labeled as 0),so I want to use FCN to learn the non-linear algorithm and identify two parts in labeled images. I have some trouble in how can I do to solve this problem, I think I should convert my data to "imdb.mat" to feed fcntrain firstly, but I didn't achieve it .Then I think I should run fcnInitializeModel to initialize the net and modify some parameters in the code.What I think is right?I hope someone can help me ,and I am grateful if you can share your code to me

Have you solved the problem？ Can you share your code to me？ Thanks in advance！
Best wishes.