what does the result image represent for

[gpcv-liujin]

When run your program, i got some result image(png format),but the result image differ greatly from the true disparity image. I tried to output the values of final_result[k] which comes from result = np.squeeze(final_result[k]),but still found the values differ greatly from the true disparities. Can you tell me what does the result image represent for ? Did i get wrong result? my result image

[ZhijianJiang]

I tried the dataset mentioned in README.md and the output is just like the result in paper.

Since I do not know about your work and dataset much, I am unable to figure out the problem.

[sunnyHelen]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

[gpcv-liujin]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

I guess the input image size（height and width） must be divisible by 32. I have fixed it for a long time，so I am not sure what else has changed. You can try it.

[sunnyHelen]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

I guess the input image size（height and width） must be divisible by 32. I have fixed it for a long time，so I am not sure what else has changed. You can try it.

Ok, thank you. I used kitti dataset. And did you use the similar pkl file "gopro.pkl" in this provided coding ? I think it 's the ground truth data, right?

[gpcv-liujin]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

I guess the input image size（height and width） must be divisible by 32. I have fixed it for a long time，so I am not sure what else has changed. You can try it.

Ok, thank you. I used kitti dataset. And did you use the similar pkl file "gopro.pkl" in this provided coding ? I think it 's the ground truth data, right?

Yes，it ‘s the ground truth data , you can get it by running the prepfm.py .

[sunnyHelen]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

I guess the input image size（height and width） must be divisible by 32. I have fixed it for a long time，so I am not sure what else has changed. You can try it.

Ok, thank you. I used kitti dataset. And did you use the similar pkl file "gopro.pkl" in this provided coding ? I think it 's the ground truth data, right?

Yes，it ‘s the ground truth data , you can get it by running the prepfm.py .

wow, so helpful. Thank you. Could you please leave me a email address? Maybe I need your help later. I'm afraid you won't see my question at Github. Please.

[gpcv-liujin]

Hey, I get almost the same result with you. I don't know why. Have you already fix it? @Jan-jin

I guess the input image size（height and width） must be divisible by 32. I have fixed it for a long time，so I am not sure what else has changed. You can try it.

Ok, thank you. I used kitti dataset. And did you use the similar pkl file "gopro.pkl" in this provided coding ? I think it 's the ground truth data, right?

Yes，it ‘s the ground truth data , you can get it by running the prepfm.py .

wow, so helpful. Thank you. Could you please leave me a email address? Maybe I need your help later. I'm afraid you won't see my question at Github. Please.

My email is [email protected]

[jiafengshen]

不知道你是怎么跑通他的程序的 @Jan-jin @sunnyHelen

[candlecove-nju]

When run your program, i got some result image(png format),but the result image differ greatly from the true disparity image. I tried to output the values of final_result[k] which comes from result = np.squeeze(final_result[k]),but still found the values differ greatly from the true disparities. Can you tell me what does the result image represent for ? Did i get wrong result?

my result image

add cmap='gray' to imsave( ), that works.