The project developed using TensorFlow to detect the no of objects every second entering building gate. work in progress

• Tensorflow
  
• openCV
  
• you must have pip and python pre installed if not install them first.

# For CPU
pip install tensorflow
# For GPU
pip install tensorflow-gpu

For Ubuntu

sudo apt-get install protobuf-compiler

For OSX

brew install protobuf

Other Libraries

pip install opencv-python

Tensorflow object detection API

protoc utils/*.proto --python_out=.

If you want to test out the implementation then you can use the person_detection_every_sec.py which is working on pre recorded video

python person_detection_every_sec.py

To run the server

python server.py

As per the original implementation of the tensorflow object detection API, the bounding boxes are normalised. To get the original dimensions you need to do the following.

(left, right, top, bottom) = (xmin * im_width, xmax * im_width,
                              ymin * im_height, ymax * im_height)

• Download and use, no license needed*