In week 7, we explored the raspberry pi grove kit given, as well as a Microsoft AI camera. We familiarised ourselves with the components given in the toolkit such as the various sen
sors and they can be connected to the main raspberry pi motherboard. We created a demo program that takes in values from a humidity sensor in regular intervals using Python to get an idea of what a subsystem code may look like.
With the AI camera, we had a look at how to connect it to our laptops and researched how we can use it to recognise objects with a trained model on azure cloud. We had a look into Custom Vision, where we can train a model by simply uploading labelled images.
In week 8, we organised a couple of meetings to brainstorm all the ideas we have that can be implemented with the resources given. We ended up with two ideas:
1. Farm Data Analysis System
sensors collect data from Farm
data sent to raspberry pi
calculations are made
raspberry pi produces an output
output sent to actuators
actuators carry out response action
2. Digital Shepherd
use data online to train a model that recognises, classifies and counts animals (Azure cloud)
deploy machine learning algorithms in ai camera
this can be incorporated in a drone to be a digital shepherd
We then presented these two ideas to our client in our next meeting. He was interested in both ideas, but ultimately preferred the first idea of having a Farm Data Analysis System with plants. However, since the second idea has a lot of potentials, our client suggests we incorporate some aspects of the AI camera into the first idea so we can still deploy some machine learning algorithms.
Comments