This content originally appeared on DEV Community and was authored by Yash Jivani
Have you ever thought of using AI to sort your customer's waste? There are many companies & waste management contractors in the market globally who want to maximize recycling and save the environment. But, the main problem is people tend to throw the waste in the bin they want or nearest to them, regardless of waste signs & symbols located on bins. Some of the key reasons are people:
- are unaware of the categories (usually there is a variation of categories)
- don't know which waste is classified under what category
- don't have time to think/wait
- and many more...
Due to the unsorted waste collection at the source, it becomes difficult for the recycling facility to sort 100% accurately. This blog shows the procedure of setting up the system (hardware & software) at/near waste collection bins.
Hardware Requirements
- Raspberry Pi
- Camera
- Arduino Uno
- Breadboard
- Wires & LED Lights
Software Requirements
- Python
- Arduino IDE
Steps.
- Configure Raspberry Pi Setup - Getting started - Raspberry Pi Documentation
- Install Python if not installed.
- SSH your Raspberry Pi into your local system or connect with HDMI display, keyboard & mouse to set up the system. This is only required once
- Create a Project Directory. Create Python Virtual Env.
- Install packages required for running the inference. You can run the trained model Inference. This document is not meant to explain Ultralytics YOLO model training & various ways of optimizing inferences for your custom use-cases & requirements using Roboflow or Ultralytics HUB Inference . The Waste Category Predictor (AI Model) predicts the waste type and sends the signal to Arduino.
- Install Arduino IDE in Raspberry Pi.
- Connect your Arduino with Baseboard & Setup the led lights. Based on the custom blinking you want for each port, Send the Code to the Arduino system. Remember the custom serial constant you set for each LED light.
- Install pyserial package and connect your app with LED using serial commands.
- Now, you are all set, once you connect the camera with the Raspberry Pie, Run the Python App.
Workflow
- The user approaches the bin to dispose of the waste item.
- When the user points the waste item to the camera, then live video frames are sent to the Raspberry where the Python Code is running.
- AI model predicts the waste type & categorizes it.
- Based on the classified waste, the response is sent back to the LED light (installed on top of the bin) so the user can get guided where s/he has to throw the waste based on the Light color.
Generally, Raspberry Pi has less computing power affecting the prediction. To tackle this problem, Connect Coral Edge TPU with Raspberry Pi if you want to run locally. Otherwise, you can connect your camera live streams to the remote server (running either in GCP/AWS/Azure or any other cloud computing platform) using a WiFi connection (Make sure you set up proper authentication, making your system secure).
This blog doesn't provide Python code & Arduino js code. But if you want to learn more, feel free to connect with me. I would help you set up the configuration & happy to work on building projects like these.
This content originally appeared on DEV Community and was authored by Yash Jivani
Yash Jivani | Sciencx (2024-08-30T03:08:55+00:00) Smart Waste Solutions: Integrating AI in Next-Gen Waste Sorting. Retrieved from https://www.scien.cx/2024/08/30/smart-waste-solutions-integrating-ai-in-next-gen-waste-sorting/
Please log in to upload a file.
There are no updates yet.
Click the Upload button above to add an update.