Inspiration
The motivation behind this project stems from the need to assist farmers and agricultural experts in maintaining the health and quality of apple orchards. Monitoring apple ripeness and detecting diseases early can significantly reduce crop losses, increase yield quality, and promote sustainable farming. The project aims to make these tasks more accessible and efficient through the use of AI technology.
What it does
The project is divided into two primary modules:
1.Apple Detection: This module uses a webcam feed to analyze apples in real-time, classifying them as ripe, unripe, or overripe while also providing a count of each category. This helps farmers decide the optimal time for harvest.
2.Apple Leaf Health Detection: This module examines apple leaf images to determine if they are healthy or affected by scab disease. The goal is to facilitate early disease detection, allowing for timely interventions.
How we built it
The project was developed using a combination of computer vision and deep learning techniques:
Data Collection: The dataset for both apple ripeness and apple leaf health was sourced from Kaggle, where high-quality images were gathered and preprocessed for model training.
Model Training: The detection models were built using the yolov10.pt model, a YOLO (You Only Look Once) variant optimized for detecting apple ripeness and identifying scab diseases in leaves.
Development: Python was the primary language used, leveraging libraries such as OpenCV for real-time image processing and analysis.
User Interface: Scripts were designed for easy interaction, allowing users to change image paths to analyze specific apple or leaf images, with ripeness detection supported via webcam input.
Challenges we ran into
Data Quality: Acquiring high-quality datasets for both apple ripeness and apple leaf health was a challenge. Ensuring diversity in the dataset for better accuracy required careful curation.
Model Accuracy: Fine-tuning the models to reduce false positives and improve the accuracy of both ripeness and disease detection took considerable effort.
Real-time Analysis: Implementing real-time detection with accurate results and minimal latency posed a technical challenge, especially for larger images.
Accomplishments that we're proud of
Successfully developing an AI-based system capable of identifying apple ripeness and leaf health with a high degree of accuracy.
Creating a user-friendly interface where farmers and agricultural experts can easily input images and receive actionable results.
Reducing the dependency on manual observation, making the monitoring process faster and more efficient.
What we learned
The importance of high-quality data for training machine learning models.
Techniques for optimizing computer vision models for specific agricultural applications.
How to build and structure an AI project to cater to both usability and precision.
What's next for AI-powered Apple Tree Health Detection
Integration with Drones: Incorporating drone-based image capture for automated and wide-scale orchard monitoring.
Mobile App Development: Creating a mobile application to make the system more accessible for farmers in the field.
Real-time Alerts: Adding a feature for real-time alerts if unhealthy leaves or overripe apples are detected.
Expanded Disease Detection: Enhancing the model to detect a broader range of diseases affecting apple leaves and fruit.
User Feedback: Gathering feedback from farmers to refine the models and features further, ensuring the tool remains practical and beneficial.
Log in or sign up for Devpost to join the conversation.