Intro

Sample project that uses flask and opencv to pass an image from client.py to server.py and perform face and pedestrian detection. The api returns the max of faces detected or pedestrians detected.

Instructions

Setup

Use with python 3.6

pipenv --python 3.6

pipenv install -r requirements.txt

pipenv shell

Running

python server.py

python client.py

In client, change the filename to use different images.

Notes

The master branch uses opencv with haar features for face detection. It uses less dependencies but is less accurate.

The branch f/cnn_face_detection uses CNNs and has better accuracy but requires more dependencies, including tensorflow.

Haar

Pedestrians

Resources

People counter https://www.pyimagesearch.com/2018/08/13/opencv-people-counter/