GiovanniPasq / DA-Faster-RCNN

Detectron2 implementation of DA-Faster R-CNN, Domain Adaptive Faster R-CNN for Object Detection in the Wild
MIT License
53 stars 10 forks source link

the installation of Detectron2 did not complete successfully #4

Closed TahereSaidi closed 1 year ago

TahereSaidi commented 1 year ago

Hello. thanks for your work! Can we use this model when we don't have a labels in the target domain?

GiovanniPasq commented 1 year ago

Hello, I tried to install detectron2 using the instructions inside my notebook and the installation was successful, it takes a couple of minutes.

Yes you can use the model when you do not have the labels in the target domain. Add this line cfg_target.DATALOADER.FILTER_EMPTY_ANNOTATIONS = False to the script after the instruction cfg_target.INPUT.MIN_SIZE_TRAIN =(600,)

TahereSaidi commented 1 year ago

When I run this command: !python -m pip install pyyaml==5.1 !python -m pip install 'git+https://github.com/facebookresearch/detectron2.git' exit(0) the file comes to this address: [/usr/local/lib/python3.7/dist-packages/detectron2/modeling/meta_arch] But my question is, is it possible to make the file to be saved in Google Drive instead of going to this address? So that after I apply the mentioned changes, it will be saved and won't be lost by closing the colab?

GiovanniPasq commented 1 year ago

Unfortunatelly you cannot change the files destination, the changes won't be saved when you close the colab :(

TahereSaidi commented 1 year ago

Thank you for your attention. My last question is that when I run these lines, I have Python 3.10 at this address and the detectron2 comes here, is there any problem that Python 3.7 is not created for me?

GiovanniPasq commented 1 year ago

No it's not a problem, the python version can change but you need to replace the file according to your installed version. Your path should be ../usr/local/lib/python3.10/dist-packages/detectron2

TahereSaidi commented 1 year ago

yes. thank you so much.

GiovanniPasq commented 1 year ago

You are welcome, i am closing the issue, if you have other problem reopen it. If you could leave a star it can help to grow up the repo :)