Unofficial python API for DeepStack. Provides classes for making requests to the object detection & face detection/recognition endpoints. Also includes some helper functions for processing the results. See the Jupyter notebooks for usage.

Run deepstack with all three endpoints active (CPU, noAVX mode):

docker run \
      -e VISION-SCENE=True \
      -e VISION-DETECTION=True \
      -e VISION-FACE=True \
      -v localstorage:/datastore \
      -p 5000:5000 \
      -e API-KEY="" \
      --name deepstack deepquestai/deepstack:noavx

Check deepstack is running using curl (from root of this repo):

curl -X POST -F image=@tests/images/test-image3.jpg 'http://localhost:5000/v1/vision/detection'

If all goes well you should see the following returned:

{"success":true,"predictions":[{"confidence":0.9998661,"label":"person","y_min":0,"x_min":258,"y_max":676,"x_max":485},{"confidence":0.9996547,"label":"person","y_min":0,"x_min":405,"y_max":652,"x_max":639},{"confidence":0.99745613,"label":"dog","y_min":311,"x_min":624,"y_max":591,"x_max":825}]}

• Create venv -> python3.7 -m venv venv
• Use venv -> source venv/bin/activate
• pip3 install -r requirements.txt and pip3 install -r requirements-dev.txt
• Run tests with venv/bin/pytest tests/*
• Black format with venv/bin/black deepstack/core.py and venv/bin/black tests/test_deepstack.py

• Docs are created using Jupyter notebooks
• Install in venv with -> pip3 install jupyterlab
• Run -> venv/bin/jupyter lab

• Generate requirements with -> pip3 freeze > requirements.txt
• Create a distribution with -> python3 setup.py sdist (creates dist)
• Upload packages with twine with -> twine upload dist/*