Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Adds customization options to the segmentation models and other improvements #263

Merged
merged 48 commits into from
Jan 28, 2021
Merged

Adds customization options to the segmentation models and other improvements #263

merged 48 commits into from
Jan 28, 2021

Commits on Jan 17, 2021

  1. adds ability to add custom callbacks

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    742d7a0 View commit details
    Browse the repository at this point in the history

  2. added tensorflow to requirements.txt

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    db326f2 View commit details
    Browse the repository at this point in the history

  3. uses model.fit instead of model.fit_generator 

    this is the newer version of fit_generator and should be used instead
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    1cae170 View commit details
    Browse the repository at this point in the history

  4. uses the newer SegmentationMapsOnImage instead of the depricated Segm… 

    …entationMapOnImage

removed now unecessary nb_classes and replaced get_arr_int(), with get_arr()
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    9291ea1 View commit details
    Browse the repository at this point in the history

  5. uses the more versitile and robust tensorflow model checkpoint

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    4b85ca4 View commit details
    Browse the repository at this point in the history

  6. uses tensorflow's simpler lastest_checkpoint function

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    fb59ebc View commit details
    Browse the repository at this point in the history

  7. adds epoch identifier in checkpoint name

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    8df1df8 View commit details
    Browse the repository at this point in the history

  8. if the checkpoint base directory does not exists create it

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    c56312e View commit details
    Browse the repository at this point in the history

  9. must use the directory for tf.train.latest_checkpoint

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    d2c7d06 View commit details
    Browse the repository at this point in the history

  10. added the ability to use custom augment functions 

    just need to pass in the function in the cusomt_augmentation argument
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    86f1afc View commit details
    Browse the repository at this point in the history

  11. replace deprecated augmentation functions

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    d1d3935 View commit details
    Browse the repository at this point in the history

  12. fixed auto resume checkpoint epoch numbering

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    9ca331e View commit details
    Browse the repository at this point in the history

  13. Added fail_safe functionality back to find_latest_checkpoint

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    929ac0e View commit details
    Browse the repository at this point in the history

  14. added legacy code back for find latest checkpoint if tf.train does no… 

    …t work
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    44cafbb View commit details
    Browse the repository at this point in the history

  15. Revert "replace deprecated augmentation functions" 

    This reverts commit d1d3935.
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    dbecfc0 View commit details
    Browse the repository at this point in the history

  16. updates the base library versions 

    updates imageio and imgaug
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    4cc835a View commit details
    Browse the repository at this point in the history

  17. made makedirs backwards compatible with Python 2.7 

    (even though there should be no need to)
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    5e556e1 View commit details
    Browse the repository at this point in the history

  18. Revert "Revert "replace deprecated augmentation functions"" 

    This reverts commit dbecfc0.
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    26a6c19 View commit details
    Browse the repository at this point in the history

  19. reverted imageio dependency and mand imagaug >=

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    ad140fb View commit details
    Browse the repository at this point in the history

  20. updates tensorflow base dependency

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    7dea6be View commit details
    Browse the repository at this point in the history

  21. Revert "updates tensorflow base dependency" 

    This reverts commit 7dea6be.
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    7fcb04e View commit details
    Browse the repository at this point in the history

  22. added python 2 backwards compatiblity for checkpoint callback 

    (though this should not be needed)
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    d0eee0e View commit details
    Browse the repository at this point in the history

  23. updated missed deprecated augmentation function

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    fae9938 View commit details
    Browse the repository at this point in the history

  24. adds the ability when vizualizing dataset to set the augmentation typ… 

    …e and image size
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    4d456a5 View commit details
    Browse the repository at this point in the history

  25. added expect_partial to remove annoying warning messages 

    these messages appear due to loading checkpoints but not using all the data
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    a478ca5 View commit details
    Browse the repository at this point in the history

  26. fixed cases where load_weights return None 

    (Python 2 again...)
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    ffcf999 View commit details
    Browse the repository at this point in the history

  27. gave pairs from paths the ability to have additional multi input func… 

    …tionality
    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    807daa7 View commit details
    Browse the repository at this point in the history

  28. added multi input data generator/augmentation

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    184d7d6 View commit details
    Browse the repository at this point in the history

  29. python 2.7 compatability list unpacking

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    5f7052d View commit details
    Browse the repository at this point in the history

  30. added to the train parameters the other inputs path options

    @Marius-Juston
    Marius-Juston committed Jan 17, 2021
    Configuration menu
    Copy the full SHA
    13c054f View commit details
    Browse the repository at this point in the history

Commits on Jan 18, 2021

  1. Added ability to add custom image preprocessing

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    76bf079 View commit details
    Browse the repository at this point in the history

  2. python 2.7 compatability import statement

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    451396e View commit details
    Browse the repository at this point in the history

  3. Added custom augmenttion data generation test

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    817606f View commit details
    Browse the repository at this point in the history

  4. moved test image generation to setup and created custom augmention wi… 

    …th multi images test
    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    26b876a View commit details
    Browse the repository at this point in the history

  5. fixed missing preprocessing step for single input data generation

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    28b8ad0 View commit details
    Browse the repository at this point in the history

  6. implemented a preprocessing datageneratio test

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    fd1a6f0 View commit details
    Browse the repository at this point in the history

  7. added preprocessing checks with multi input system

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    b249a43 View commit details
    Browse the repository at this point in the history

  8. added smple data generator multi input test

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    5f39a53 View commit details
    Browse the repository at this point in the history

  9. made testing with multi input data generation more complex 

    added new folder and new test images for testing harder
    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    147a440 View commit details
    Browse the repository at this point in the history

  10. removed type hinting (Python 2.7 compatability)

    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    e488684 View commit details
    Browse the repository at this point in the history

  11. fixed python 2 infinite loop 

    in python 2.7 itertools.cycle seesm to give an infinite loop, so removed convoluted cycle and just use image seg pairs to get things to work
    @Marius-Juston
    Marius-Juston committed Jan 18, 2021
    Configuration menu
    Copy the full SHA
    07f0144 View commit details
    Browse the repository at this point in the history

Commits on Jan 20, 2021

  1. added ability to choose tensorflow input image type

    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    436e874 View commit details
    Browse the repository at this point in the history

  2. Added ability to change input channel size

    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    0892b44 View commit details
    Browse the repository at this point in the history

  3. python 3 to 2 import compatability

    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    f732ed4 View commit details
    Browse the repository at this point in the history

  4. Added failsafe when loading from imagenet pretrained network 

    in case the first layer was modified
    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    fffc021 View commit details
    Browse the repository at this point in the history

  5. for sub mean added fail safe in case image has more or less than 3 ch… 

    …annels
    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    1 Configuration menu
    Copy the full SHA
    f45a368 View commit details
    Browse the repository at this point in the history

  6. fixed fcn_mobilenet 224 input height and width requirements

    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    df8c1a0 View commit details
    Browse the repository at this point in the history

  7. added missing image type parameters

    @Marius-Juston
    Marius-Juston committed Jan 20, 2021
    Configuration menu
    Copy the full SHA
    c961543 View commit details
    Browse the repository at this point in the history