Detection of Animals in Wildlife with Video Cameras

German Suvorov, PhD
Product AI
Published in
1 min readApr 8, 2021

Automatically detect wildlife in cameras despite remote locations and poor internet connection.

Solution by: Data Monsters

Date: March 2021

Challenge

In the wild, video cameras are installed to monitor the environment. Most of the time, nothing happens in their field of vision. The task is to automatically detect the moments when an animal appears in the camera’s field of view, determine what kind of animal it is, and send a notification and a video clip to subscribers. Cameras are installed in places with a poor internet connection, so calculations must be performed on the edge.

Solution

The signal from the video camera goes to NVIDIA Jetson Nano and is split into frames, after which common animals such as deer, elk, fox, and others are detected in real-time. If an animal is found, the message is sent to subscribers using the Firebase Cloud Messaging platform. Rebroadcasting of video from cameras is carried out using the Dacast streaming video platform.

Technologies used

  • NVIDIA Jetson Nano
  • NVIDIA DeepStream SDK
  • Tensor RT

--

--

German Suvorov, PhD
Product AI

Industrial AI solution architect and engineer. German’s background is in automotive manufacturing, manufacturing automation, supply chain management, AI.