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

Feat/unfreeze layers fpn backbone #2160

Merged
merged 17 commits into from May 15, 2020
Merged

Feat/unfreeze layers fpn backbone #2160

merged 17 commits into from May 15, 2020

Commits on Apr 29, 2020

  1. freeze layers only if pretrained backbone is used 

    If pretrained backbone is not used and one intends to train the entire network from scratch, no layers should be frozen.
    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    837bed5 View commit details
    Browse the repository at this point in the history

  2. function argument to control the trainable features 

    Depending on the size of dataset one might want to control the number of tunable parameters in the backbone, and this parameter in hyper parameter optimization for the dataset. It would be nice to have this function support this.
    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    c57c66a View commit details
    Browse the repository at this point in the history

  3. ensuring tunable layer argument is valid

    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    e972118 View commit details
    Browse the repository at this point in the history

  4. backbone freezing in fasterrcnn_resnet50_fpn 

    Handle backbone freezing in fasterrcnn_resnet50_fpn function rather than the resnet_fpn_backbone function that it uses to get the backbone.
    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    44e94ba View commit details
    Browse the repository at this point in the history

  5. remove layer freezing code 

    layer freezing code has been moved to fasterrcnn_resnet50_fpn function that consumes resnet_fpn_backbone function.
    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    4aa9f04 View commit details
    Browse the repository at this point in the history

  6. correcting linting errors

    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    1f79756 View commit details
    Browse the repository at this point in the history

  7. correcting linting errors

    @muaz-urwa
    muaz-urwa committed Apr 29, 2020
    Configuration menu
    Copy the full SHA
    f93f80b View commit details
    Browse the repository at this point in the history

Commits on Apr 30, 2020

  1. move freezing logic to resnet_fpn_backbone 

    Moved layer freezing logic to resnet_fpn_backbone with an additional parameter.
    @muaz-urwa
    muaz-urwa committed Apr 30, 2020
    Configuration menu
    Copy the full SHA
    e1c9592 View commit details
    Browse the repository at this point in the history

  2. remove layer freezing from fasterrcnn_resnet50_fpn 

    Layer freezing logic has been moved to resnet_fpn_backbone. This function only ensures that the all layers are made trainable if pretrained models are not used.
    @muaz-urwa
    muaz-urwa committed Apr 30, 2020
    Configuration menu
    Copy the full SHA
    abd3707 View commit details
    Browse the repository at this point in the history

  3. update example resnet_fpn_backbone docs

    @muaz-urwa
    muaz-urwa committed Apr 30, 2020
    Configuration menu
    Copy the full SHA
    3ddfef0 View commit details
    Browse the repository at this point in the history

  4. correct typo in var name

    @muaz-urwa
    muaz-urwa committed Apr 30, 2020
    Configuration menu
    Copy the full SHA
    2ef1497 View commit details
    Browse the repository at this point in the history

  5. correct indentation

    @muaz-urwa
    muaz-urwa committed Apr 30, 2020
    Configuration menu
    Copy the full SHA
    88b0439 View commit details
    Browse the repository at this point in the history

Commits on May 6, 2020

  1. adding test case for layer freezing in faster rcnn 

    This PR adds functionality to specify the number of trainable layers while initializing the faster rcnn using fasterrcnn_resnet50_fpn function. This commits adds a test case to test this functionality.
    @muaz-urwa
    muaz-urwa committed May 6, 2020
    Configuration menu
    Copy the full SHA
    c40958a View commit details
    Browse the repository at this point in the history

  2. updating layer freezing condition for clarity 

    More information in PR
    @muaz-urwa
    muaz-urwa committed May 6, 2020
    Configuration menu
    Copy the full SHA
    24e77c7 View commit details
    Browse the repository at this point in the history

  3. remove linting errors

    @muaz-urwa
    muaz-urwa committed May 6, 2020
    Configuration menu
    Copy the full SHA
    ca8b180 View commit details
    Browse the repository at this point in the history

  4. removing linting errors

    @muaz-urwa
    muaz-urwa committed May 6, 2020
    Configuration menu
    Copy the full SHA
    9045110 View commit details
    Browse the repository at this point in the history

  5. removing linting errors

    @muaz-urwa
    muaz-urwa committed May 6, 2020
    Configuration menu
    Copy the full SHA
    552221b View commit details
    Browse the repository at this point in the history