Introduction

This project aims to reduce Human-wildlife conflict with the help of video surveillance and artificial intelligence. The idea is to use a camera for 24h video surveillance and machine learning to detect and alert for animals that pose a threat livelihood or safety of the people in the area.

Structure

Capture app stream - captures and saves photos from a cctv camera Capture app - captures and saves photos from the camera connected to RaspberryPi Motion app - performs motion detection over a given perimeter Classify app - using neural network trained on Edge Impulse to detect objects in images from the input folder. Saves the images in which the objects of interest were found on "detected" folder Upload Edge - upload images from "detected" folder to Edge Impulse Addnotation Queue Upload Telegram - upload images from "detected" folder to Telegram group

How to install

Install Python >3.9 Install requirements

pip install -r requirements.txt

Install Edge Impulse CLI

How to run

python picamera.py Config.json

python motion.py Config.json

python classify-image.py Config.json

python upload_edge.py Config.json

python upload_telegram.py Config.json