-
Notifications
You must be signed in to change notification settings - Fork 7.4k
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.
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
[GSoC] TF 2.0 Mask R-CNN #78
base: master
Are you sure you want to change the base?
Conversation
Vishal-V
commented
Jul 22, 2019
- TF 2.0 compatible Mask R-CNN model adapted from the Matterport implementation.
- Performance improvements with lesser parameters and improved residual blocks.
- Parallelized training with GPUs and an assortment of utils for various use cases.
- Associated Migration guide for the Tensorflow 2.0 specific changes.
- Runner script to run the model and draw inferences form the pre-trained COCO weights.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks great, I'll need to go through it in more detail.
Two top level comments:
-We need to make sure we handle the "adapted from" correctly.
-(optional) It would be nice if you had a notebook version of the driver script.
|
||
Reference: | ||
[Mask R-CNN](https://arxiv.org/asb/1703.06870) | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These "adapted form" make me a little nervous.
I'll need to check and make sure we attribute this properly.
text = "{}: {:.3f}".format(label, score) | ||
y = startY - 10 if startY - 10 > 10 else startY + 10 | ||
cv2.putText(image, text, (startX, y), cv2.FONT_HERSHEY_SIMPLEX, 0.6, color, 2) | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A driver script should use if __name__=="__main__":
# Utility Functions | ||
############################################################ | ||
|
||
def log(text, array=None): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This file is huge. Can you break this up?
For example, putting utility functions in utils.py, model in models.py, training in train.py, etc
Also, is it possible to add MirroredStrategy (https://www.tensorflow.org/versions/r2.0/api_docs/python/tf/distribute/MirroredStrategy) to this model?