DeepGlobe Road Extraction Challenge Forum

Go back to competition Back to thread list Post in this thread

> Dataset Request

Dear Admin,
Could you approve my request for road extraction dataset, Thank you.

Posted by: canaltinigne @ Aug. 5, 2019, 8:22 a.m.

Approved.

Posted by: ilkedemir @ Aug. 8, 2019, 11:44 p.m.

Hi, I noticed that you submitted your result successfully. However, I got some problem submitting my results.

I made the zip files with my results stored in the root directory and all PNGs are named with the format as 'num_mask.png' and valued in 3 channels with 0 and 255. I also tried the value of 0 and 1. However, it came out with the same errors.

Output log showed '===== Some files are missing in upload ===== '.

Error log was showed as follow without ERRORS but only WARNING:
WARNING: Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap.
/opt/conda/lib/python2.7/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.
warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')

I even tried rename PNGs in train set with the name list of test set and submit.The result is the same.

Please help! Thanks a lot!

Posted by: xhuang @ Sept. 10, 2019, 1:51 a.m.

Are you sure that you submit the predictions for 1243 test images ?

Posted by: canaltinigne @ Sept. 10, 2019, 7:32 a.m.

emm, thank you for replying.
I downloaded the road_test_sat.zip from Phase #2 Final and got 1101 pictures. I used them to generate predictions and uploaded .
I tried val set with 1243 prediction images you mentioned, same problem...
I searched for the warning. It's related to docker configuration...
I don't really know what was wrong with my process or was there anything wrong with the sever after your testing?
emm...Here is a little excessive ask. Could you send one single image with your upload format to me?
If you are willing to help, my email address is meiguidezang@qq.com.

Thanks a lot!

Posted by: xhuang @ Sept. 10, 2019, 8:25 a.m.
Post in this thread