fonte da imagem: link

!wget https://raw.githubusercontent.com/awarischool/br-data-science/master/image-colorizer/deoldify_wrapper.py

--2020-08-09 23:02:30--  https://raw.githubusercontent.com/awarischool/br-data-science/master/image-colorizer/deoldify_wrapper.py
Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...
Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 2614 (2.6K) [text/plain]
Saving to: ‘deoldify_wrapper.py’

deoldify_wrapper.py 100%[===================>]   2.55K  --.-KB/s    in 0s      

2020-08-09 23:02:31 (45.5 MB/s) - ‘deoldify_wrapper.py’ saved [2614/2614]

from deoldify_wrapper import DeOldify

Trying to import DeOldify (if installed)
No module named 'deoldify'
DeOldify not found, installing..
  Cloning DeOldify Repository...
  Opening DeOldify Folder
  Installing Colab requirements...
  Importing DeOldify

/content/deoldify_wrapper.py:42: UserWarning: WARNING: GPU not available. Activate it on Colab at Edit > Notebook Settings
  warnings.warn('WARNING: GPU not available. Activate it on Colab at Edit > Notebook Settings')

  Downloading Colorizer Model

deo = DeOldify()

  Initializing Colorizer

/usr/local/lib/python3.6/dist-packages/fastai/data_block.py:442: UserWarning: Your training set is empty. If this is by design, pass `ignore_empty=True` to remove this warning.
  warn("Your training set is empty. If this is by design, pass `ignore_empty=True` to remove this warning.")
/usr/local/lib/python3.6/dist-packages/fastai/data_block.py:445: UserWarning: Your validation set is empty. If this is by design, use `split_none()`
                 or pass `ignore_empty=True` when labelling to remove this warning.
  or pass `ignore_empty=True` when labelling to remove this warning.""")
Downloading: "https://download.pytorch.org/models/resnet101-5d3b4d8f.pth" to /root/.cache/torch/hub/checkpoints/resnet101-5d3b4d8f.pth

Done!

url = 'https://i.imgur.com/ziXRlN6.jpg'
deo.colorize(url)

foto original

Conclusão:

Como é possível verificar ficou muito bom mas ha algumas diferenças entre a foto original colorida e a foto colorizada pela aplicação ela ficou parecendo um filtro de envelhecimento de fotos que encontramos no celular.Um desses motivos foi o fato de eu ter usado o filtro do celular, um Xiaomi S2, para deixar a foto em P&B e colocado no modo filme. Deixo o desafio de deixar a foto original em preto e branco pelo fotophoshop e verificar se o resultado será mais próximo da original.