Farz is an AI waste management company and the winner of Product of the Year at Injaz Bahrain Company Program 2024. Our flagship product, Fariz-01, is an AI-powered recycling bin that uses computer vision to identify and sort different types of waste at the point of disposal.
Fariz-01 is an autonomous waste management solution designed and built in Bahrain. It leverages deep learning models to classify waste items and sort them automatically, making recycling easier and more efficient.
- AI-powered: Uses computer vision to detect and classify waste.
- Edge deployment: Runs on a Raspberry Pi for real-time, on-device inference.
- Award-winning: Product of the Year, Injaz Bahrain 2024.
For hardware and electronics details, see the Farz Electronics Repo.
The model is trained on a curated and cleaned dataset of waste images.
- Dataset: Garbage Classification 3 (Roboflow)
The model classifies waste into the following categories:
- 0: BIODEGRADABLE
- 1: CARDBOARD
- 2: GLASS
- 3: METAL
- 4: PAPER
- 5: PLASTIC
Install the following Python dependencies:
import numpy as np
from PIL import Image
import tensorflow as tf
import time
import cv2
import ultralyticsYou can install them using pip:
pip install numpy pillow tensorflow opencv-python ultralyticsThe scripts in this repository are designed for use on a Raspberry Pi or any Linux-based system with a camera.
-
YOLOv8 Inference:
Use scriptYOLO.py or script2.py to run real-time object detection using the YOLOv8 model. -
TensorFlow Lite Inference:
Use scripTFLITE.py for lightweight inference with a TFLite model.
Model training is performed using Google Colab and the Ultralytics YOLO library.
Refer to the Jupyter notebook Garbage_Classification_Yolov8.ipynb for the full training pipeline.
For hardware integration, Raspberry Pi setup, and bin actuation, refer to the Farz Electronics Repo.
This project is for educational and demonstration purposes.
- Injaz Bahrain Company Program 2024
- American University of Bahrain
- Roboflow for dataset hosting
For questions or support, please open an issue or contact the Farz team.