Open Paulito-7 opened 6 years ago
@Paulito-7 We participated as EOS DA team and got 5th place on round 1, but result is still 5 points worse then segmentation with postprocessing.
@maksimovkonstantin thank you for your reply. Can you elaborate on the post-processing you used to get such a performance? I saw interesting improvements here for post-processing Is your trained weight file available? Now that the challenge is finished, I would be curious to know how you dealt with the problem.
Thank you!
Hello everyone,
I wonder if someone has trained Mask R-CNN for the CrowdAI mapping challenge, and I am very curious about the results, yet my computer is too weak to make such an end-to-end training. Did you have good results? Can you make the trained weights available to work on improvement?
Many thanks !