Image of good and spoiled apples
Identified the good and spoiled apples

Spoiled Apple Detector Project

Introduction

I built a spoiled apple detector that identifies spoiled apples within a group, aiming to reduce food wastage by separating spoiled ones from good apples. This model addresses spoilage in apples and can be extended to other food items, broadening its impact on reducing food waste.

Inspiration

The growing problem of food wastage inspired me to create this project. With a significant amount of food being discarded due to spoilage, I wanted to find a solution that could help extend the shelf life of good food and minimize waste.

Project Overview

By fine-tuning state-of-the-art computer vision algorithms, I successfully developed a system that effectively identifies spoiled apples, helping to reduce food wastage. This project demonstrates how combining methodologies like object detection and binary classification can yield effective solutions, even when faced with data challenges.

Building the Project

Data Collection: I gathered images of both good and spoiled apples.
Binary Classifier: I developed a binary classifier by fine-tuning the 'MobileNetV2' model to distinguish between good and spoiled apples.
YOLO Integration: I fine-tuned YOLO to detect apples in the images.
Prediction Pipeline: Detected objects from YOLO were sent to the binary classifier for final prediction.

Challenges Faced

Dataset Issues: The primary challenge was the lack of a correctly annotated dataset to train YOLO effectively. I had to create my own dataset and train the binary classifier first.
Model Performance: Ensuring that both the YOLO model and the binary classifier performed well in detecting and classifying the apples was a critical aspect I focused on.

Learnings

I learned the importance of fine-tuning with a well-annotated dataset and the implications it has on model performance.
I gained hands-on experience with YOLO and fine tuning techniques, enhancing my skills in computer vision and machine learning.

Conclusion

This project not only helped me apply my skills in a practical scenario but also contributed to a cause I am passionate about—reducing food wastage. I hope to further refine the project and expand its applications to other types of food items.

Spoiled Apple Detector