Vision algorithm - SEMS

This repository aim to run a script (main.py) to process n videos/streams.

• Algorithms implemented:
  • People Counter
  • People Tracker
  • 2D Distance Violation
• Uses Flask to publish the image post-processed into a web-page to localhost:8080.
• Communicates directly with SEMS backend to request the source of the videos/streams and to publish all vision data calculated.

It also aims to run a script using ros (roslaunch sems_vision covid19_measures.py) to process the video stream of a zed2 camera.

• Algorithms implemented:
  • Mask Usage
  • 3D Distance Violation

The code is tested using Ubuntu 18.

System Requirements

• python3
• python-pip
• virtualenv (Recommended)

Python Requirements

Virtualenv

• Create

If you haven't create a virtualenv, create one.

virtualenv -p python3 _NAME_

• Activate

Before running the script active the virtualenv.

source _NAME_/bin/activate

• Deactivate

If you are done running the script deactive the virtualenv.

deactivate

Dependencies

Use the package manager pip to install the requirements after you have activated your venv created in somewhere else.

pip install -r requirements.txt

Usage script

Open main.py, modify global variables if needed.

CAMARAIDS = [6, 7]
BACK_ENDPOINT = ["http://sems.back.ngrok.io/", "http://localhost:3001/"][0]
NGROK_AVAILABLE = True
GPU_AVAILABLE = True
VERBOSE = False
CONFIDENCE_ = 0.3
SKIP_FRAMES_ = 25

Requirements:

• Valid CamaraIDS.
• Backend running.
• Yolov3.
• Python Dependencies.
• Videos Folder.

Run main.py

Open localhost:8080, all camaras should be displayed overthere.