Capstone Project for Data Analytics for Business at St. Clair College
One of the main challenges that we face when we arrive in Canada as international students is finding a job, recycling is one of the ways that we can earn some extra money without uploading hundreds of resumes or hacking a personal interview, and the best part is we are doing great work for our new city and our planet.
Currently, the city of Windsor's Open Data Catalogue records nearly 5,000 service requests for uncollected recycling in 2022. Assuming an average of 5 glass bottles (10¢), and 15 cans (TBS/LCBO) (10¢), a basic profit of $10,000 can be generated.
However, monetary gain is not the sole focus. The Government of Canada emphasizes the importance of recycling:
"Canadians throw away over 3 million tonnes of plastic waste every year. Only 9% is recycled while the rest ends up in our landfills"
This project aims to enhance the circular life of recycled materials in Windsor by engaging St. Clair students in the process.
A key project outcome entails the development of a mobile application that showcases current recycling collection requests in the area. By leveraging machine learning algorithms, we can optimize collection routes to maximize the number of recyclable materials collected, resulting in greater financial returns for participants and considering factors such as waste weight and type (e.g., glass bottles, cans, etc.).
In essence, this initiative operates similarly to business models such as Uber or Lyft, but instead of connecting riders with drivers, we connect recyclable object collectors with multiple recycling service requests.
App UI
- Request tab: In this tab, people can upload a picture taken with some of the recycling objects that the model was trained to detect (tin cans, glass bottles, plastic bottles). The interface will show then the number of elements per class together with the latitude and longitude where the image was taken.
- Map: This tab shows all the requests collected on a map. I have been collecting images in Windsor for around four months, and so far I have taken more than 200 pictures with one or more of the targeted objects.
- Dashboard: This tab shows a dashboard about the key metrics for this project. It has the total number of requests, the total cans, glass bottles, and plastic bottles detected among all the requests. In addition, there is a bar chart with the number of requests per day since I started collecting images. Lastly, there are three heatmaps that visualize the magnitude of the presence of each of the classes in Windsor areas.
A web application that leverages AI-powered object detection and data analytics to make recycling easier and more efficient. This app identifies recyclable items like cans, glass bottles, and plastic bottles from images and allows users to submit collection requests via an interactive map. 🌟
Key Features:
✅ AI-powered object detection using a custom-trained YOLOv5 model
✅ Interactive geospatial mapping with Pydeck and ipyleaflet
✅ Dynamic dashboard visualizations with Plotly, showcasing trends and KPIs
✅ Data storage and management with PostgreSQL
✅ Reverse geocoding for extracting geolocation insights
This project combines data analytics, computer vision, and geospatial analysis to create a seamless user experience. Here are some of the key data analytics concepts and tools applied:
- Data filtering and aggregation for trend analysis
- Interactive dashboards to visualize key performance indicators (KPIs)
- Geospatial data visualization to map recycling requests and identify patterns
Walkthrough Video to see the app in action:
