Create YOLACT++ Google Colaboratory demo

[mrm8488]

I think this is so helpful to get people into YOLACT/YOLACT++ in the easiest way!
Just one click and everything is ready :)

[abhigoku10]

@mrm8488 where is the google colab file , is it only the inference pipeline or even training is available . Wonderful thought in getting into the colab

[mrm8488]

It's in the root dir. It's for inference.

[abhigoku10]

@mrm8488 thanks a lot i was looking into wrong directory

[breznak]

Hi @mrm8488 , this a is great work! 👍 Thanks for setting the Collab up.

I've tested and reviewed your PR,

• mostly some formatting improvement suggestions
• I got the image upload fail on me (FF, linux)
  • would be good to include some out of the box images to test (already avail in the repo)

[breznak]

can we merge the

git clone
cd yolact
mkdir weights

into one execution block?

[breznak]

I think this should be part of the section below

"# Compile deformable convolutional layers (from DCNv2) to use YOLACT++ version"

[breznak]

You might add more details from the Readme, ie the different pretrained nets and their FPS/mAP as a documentation here. Ofc only one net is downloaded

[breznak]

I'm getting an err executing this cell:

Upload widget is only available when the cell has been executed in the current browser session. Please rerun this cell to enable.

---------------------------------------------------------------------------

MessageError                              Traceback (most recent call last)

<ipython-input-12-354be791d27e> in <module>()
      1 from google.colab import files
      2 
----> 3 uploaded = files.upload()
      4 pic_names = list(uploaded.keys())
      5 for pic_name in pic_names:

2 frames

/usr/local/lib/python3.6/dist-packages/google/colab/_message.py in read_reply_from_input(message_id, timeout_sec)
    104         reply.get('colab_msg_id') == message_id):
    105       if 'error' in reply:
--> 106         raise MessageError(reply['error'])
    107       return reply.get('data', None)
    108 

MessageError: TypeError: google.colab._files is undefined

Running on FF on linux

[breznak]

...as the custom image upload may fail, offer a cell running detection on already available image(s) in the repo.

[dbolya]

I think this is a great idea! It looks like Adam Kelly has also created a collab notebook that also has video support: https://colab.research.google.com/drive/1ncRxvmNR-iTtQCscj2UFSGV8ZQX_LN0M
(Blog post here: https://www.immersivelimit.com/tutorials/yolact-with-google-colab)

Should we merge the functionality of both notebooks?

[breznak]

I'm not sure if the author of this PR is still active, @mrm8488 ?

Anyway, the Adam's notebook worked great for me!

Adam Kelly has also created a collab notebook that also has video support: https://colab.research.google.com/drive/1ncRxvmNR-iTtQCscj2UFSGV8ZQX_LN0M
(Blog post here: https://www.immersivelimit.com/tutorials/yolact-with-google-colab)

• is there an OSS licence to the file?
• can we add it to the repo? (so we can make modifications)

[dbolya]

@breznak He reached out to me about the existence of it, so I can ask him. Worst case we can just link it right? Or is it better to have it in the repo (in which case we can roll our own / update this one).

[mrm8488]

Ofc, you can do you decide with my PR. License is the same as this repo.

[breznak]

Worst case we can just link it right?

true, we can always put a link in the Readme.

Or is it better to have it in the repo (in which case we can roll our own / update this one).

imho slightly better:

• we can manage updates/fixes to the file
• google warns that "you are executing a 3rd party ipython notebook from xxx" (it'd be better to have this repo as "author" rather than another nick)

[abhigoku10]

@mrm8488 @breznak @dbolya hi thanks for opensourcing and sharing the colab to run the yolact++ but when i try to do the same by myself by creating a repo and uploading all the files and changing the git clone from author to my repo and run it i get error in evaluation as mentioned below "File "/content/yolact_adas/backbone.py", line 11, in DCN
raise Exception('DCN could not be imported. If you want to use YOLACT++ models, compile DCN. Check the README for instructions.')
Exception: DCN could not be imported. If you want to use YOLACT++ models, compile DCN. Check the README for instructions." is it not possbile to do it ?? or should we perform only authors repo

[SnehashishSrivastava]

really interesting and helpful

[breznak]

@dbolya can we merge this?