search

lukemelas / deep-spectral-segmentation

[CVPR 2022] Deep Spectral Methods: A Surprisingly Strong Baseline for Unsupervised Semantic Segmentation and Localization
229 stars 41 forks source link

error in voc.py when running train.py for semantic segmentation #10

Open SimaDidari opened 2 years ago

SimaDidari commented 2 years ago

Hi Thanks for sharing your code.

  1. Running the train.py I received a bug as " init() got an unexpected keyword argument, transform_tuple, in voc.py line 159

    • Can you pls help to resolve the bug?

  2. I ran the semantic segmentation code without the train.py step for 5 times with different seeds and the average mIoU is 23.8 , The reported mIoU in paper is 30.8. Can you pls let me know if my result is in the range that you expected ? If not can you pls give me some suggestions on how to improve the result? Many thanks

chandagrover commented 1 year ago

@SimaDidari: Same issue with me for semantic segmentation. I am also getting mIoU scores for Semantic Segmentation in the ranges of ~24. Let me know if you are able to improve these scores.

SimaDidari commented 1 year ago

No still the same problem On Oct 17, 2023, at 3:23 AM, Chanda Grover @.***> wrote: @SimaDidari: Same issue with me for semantic segmentation. I am also getting mIoU scores for Semantic Segmentation in the ranges of ~24. Let me know if you are able to improve these scores.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>

chandagrover commented 9 months ago

Hi Thanks for sharing your code.

  1. Running the train.py I received a bug as " init() got an unexpected keyword argument, in voc.py line 159

    • Can you pls help to resolve the bug?

  2. I ran the semantic segmentation code without the train.py step for 5 times with different seeds and the average mIoU is 23.8 , The reported mIoU in paper is 30.8. Can you pls let me know if my result is in the range that you expected ? If not can you pls give me some suggestions on how to improve the result? Many thanks

Replace transforms_tuple with train_transform_tuple/.

I was also getting around 23 or 24 mIoU, and now I am getting around 28 or 29 without self-training, after I resized all of the images to a standard size say, 512*512.