-
Notifications
You must be signed in to change notification settings - Fork 150
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
Add timm backbones to mmdet models #848
base: master
Are you sure you want to change the base?
Conversation
Codecov Report
@@ Coverage Diff @@
## master #848 +/- ##
==========================================
- Coverage 87.62% 87.06% -0.56%
==========================================
Files 232 248 +16
Lines 4994 5367 +373
==========================================
+ Hits 4376 4673 +297
- Misses 618 694 +76
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
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.
Seems good on a high level, just a couple of things to address before we merge:
- Carefully handle
timm
imports, right now doingfrom icevision.all import *
will try to importtimm
no matter if it's installed or not - remove commented out code
- rename backbone names that start with
ice
28b2ee1
to
fc28ad6
Compare
8c06be3
to
20f7fd5
Compare
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
I added mmdet + timm support. The API is very similar to the existing one. I separate the namespaces in order to be able to navigate through the different backbones families and have the autocompletion kicking in.
Both
mobilenet
andresne(s)t
backbones plugged to aretinanet
model are provided as an example showing how a user could add support for a specific backbone/model combination. Given we have over 250 timm backbones and over 50 mmet models, it isn’t realistic to imagine a support for all backbone/model combinations. In both the documentation and the notebook, we show all the different ways to plug a Timm backbone to an MMDetection model.closes #847