Raspberry Pi ML-Based Bird Detector System

MagPi’s new edition 103 is out in the market and you can read it on the official website of Raspberry Pi. A lot of great projects are the highlights of the magazine that got attention from the tech community. One of the projects that mesmerized people the most is Mike Sadowski’s Raspberry Pi camera system for bird feeders. The idea of this system came with the necessity for the original images so that he can feed them into the machine learning system to make it efficient. In all the hustle of finding the original images, one day Mike was looking outside from the window and suddenly an inspiration sparked in his mind after looking at the bird feeder. We’ll discuss the whole working process, components, and final product in this article for a better understanding of the project. SB Components

About the Maker

Mike Sadowski, an IT executive, who has the treasure of experience of building hi-tech products that have made him big in the tech- community. He has worked as a CTO in two companies previously which provided him the experience to develop products for the users to make their life more comfortable. In an interview with Magpi, he talked about his interest in technology and the development of the products. 

Components

The project seems to be interesting and the overwhelmed response of the people has made it on the pages of Magpi. There are following components that is used in the project

  • Raspberry Pi Camera Module
  • Raspberry Pi 4 Model B
  • Amazon Web Services AWS Lambda 
  • Amazon Web Services AWS Rekognition 
  • Amazon Web Services AWS S3
  • Raspberry Pi Raspbian

Project

In this project, Raspberry Pi sends pictures from bird feeder to an S3 bucket at AWS. AWS Lambda is a computing service, provided by amazon, in which one does not have to manage servers to run codes in a process. Lambda is a cloud based software that runs the codes as per the needs which make it energy efficient and technologically advanced product. The base of the project is using the machine learning based image recognition capabilities to determine what's in the photo. If the Raspberry Pi camera detects bird or squirrel, then there are two different SNS topics, logical access points that act as a communication channel, from which the user can get the message or email. It would help the users to get the information about the detection of the birds and squirrels. If anyone wants to build this project then they need to look at the following diagram for a better understanding of the project.

Hardware for Project

You can see the Raspberry Pi camera and Raspberry Pi 4 in the picture mentioned above in the blog. A $50 telephoto lens can be used with the Raspberry Pi camera to enhance its efficiency to capture the images of the birds and squirrels. It fits on the HQ camera comfortably that makes an increment in the performance of the camera capture quality. Creator of the project used a Canon lens that he borrowed from his Canon EOS camera. One needs to buy an adapter (from Arducam) to connect a Canon lens to the camera ($30). The quality of the images comes fine with high quality and resolution. 

Motion Detection With Raspberry Pi

Raspberry Pi helps with the motion detection of the observed part like birds or squirrels. There is a package called Pi-Timolo that is a software package to run the project efficiently. You can directly run it on Raspberry Pi and in this project Pi-Timolo made the project impeccable. It helped Mike to filter out the images that become useful for him to raise the level of their product.

Leave a comment

All comments are moderated before being published