Wildlife Identification via Machine Learning
A workshop designed for students in years 11/12 which will teach technical skills, problem solving, creative use of programming, machine learning, and data stewardship. Students will create an artificial intelligence that can identify wildlife in the field. The AI model will be deployed on embedded hardware in the field. Students will customise the AI based on their location and target species. Students will learn the skills to interact directly with embedded hardware and deploy low-power, high-performance systems.
This workshop was built with support from the Google exploreCSR program and TensorFlow.
The worskhop contains 6 worksheets, each of which take about 2 hours to complete:
- Setting up the hardware
- Remote Connection
- Running AI Models
- Creating a custom model
- Running new model
- Testing new model
Details
The purpose of the workshop is to introduce:
- machine learning
- embedded systems/internet of things
- programming and linux skills
Pre-requisite knowledge is:
- basic computing skills
- interest in computers, programming, or mechatronics
Hardware required is:
- Raspberry Pi
- SD card (with custom startup image)
- Sense Hat
- CSI camera
- serial/usb cable
- PC/Mac with USBA port and web camera
- battery pack
- micro-usb cable
- WiFi access
Software Required (all freely available):
- Web Browser
- Putty(Windows)/Screen(Mac)
- CyberDuck