There are 2 parts to this project ML (the machine learning bit) and cli (the interactive bit). Currently the dataset for the machine learning section is not very accurate (~30%) because I don't have enough data. Therefore, I have decided to leave this bit out of the current build. This code will run on any machine including a raspberry pi, mac and windows computer. There is a website in the very early stages (it's just a header) found at github.com/PaddyCooper08/rf-website
These instructions are for a windows computer, you may need to change the commands to match your os.
- Clone the git repo to your local machine (this requires git to be installed on your computer):
git clone "https://github.com/PaddyCooper08/recycling-identifier"
- Create a new virtual environment to house the dependencies (this could take a while):
python3 -m venv ./venv
- Activate said virtual environment:
cd venv/scripts/ && activate
- Install the dependencies (this will take quite a long time):
pip install -r requirements.txt
(you need to be in the same directory as requirements.txt) - Build the data model:
cd ./ML
python3 global.py
Wait for this to finish then:
python3 train_test.py
You will be shown a graph with the accuracies and the model is stored in ML/output.
- cd into the correct folder:
cd cli/paddy-recycling-identifier/paddy_recycling_identifier
Command | Arguments |
---|---|
--help | n/a |
recycle | Numbers 1-7 |
See the showcase video for a better tutorial. |
- -In the works - Create website/desktop app to replace cli
- Improve dataset accuracy
- When more accurate, implement dataset