Challenge background
Agriculture plays a vital role in Bangladesh's economy, contributing 11.5% to the GDP. Fruits comprise 10% of national income. Bangladesh ranks 7th in mango production globally and it is known as the king of fruits. Bangladesh’s annual mango production is around 1.2 million metric tons from over 100,000 acres of land. However, despite its potential, mango production in the country faces challenges, including pest attacks and diseases caused by bacteria, fungi, viruses, and insects. These diseases lead to a substantial annual yield loss of around 30%, impacting farmers’ livelihoods and national production.
The problem
Bacterial and fungal diseases are major constraints for mango production, causing around 30% yield loss annually. The absence of real-time, automated systems for early detection and classification of mango leaf diseases hampers efforts to mitigate crop losses. Currently, farmers face delayed diagnoses which reduces productivity and causes financial losses.
This project aims to address this problem by developing a cutting-edge computer vision-based model that provides instant in-field detection and classification of mango leaf diseases, empowering farmers with timely information to reduce losses and enhance their income.
Goal of the project
The project goals are:
- Collect a comprehensive dataset of mango leaf images encompassing multiple bacterial and fungal diseases, ensuring representation across various regions.
- Train and optimize Convolutional Neural Network (CNN) models to accurately detect and classify mango leaf diseases using the collected dataset.
- Develop an intuitive user interface with trained models for real-time mango disease screening by farmers.
Project timeline
- 1
Week 1
- Data Collection
- Brainstorming
- Assigning task leaders
- 2
Week 2
- Data preprocessing
- Exploratory Data Analysis
- 3
Week 3
- Model training
- 4
Week 4
- Model evaluation
- 5
Week 5
- Model Deployment
What you'll learn
1. Gain hands-on experience in training CNN models using popular frameworks such as TensorFlow, applying transfer learning, and optimizing model performance.
2. Acquire knowledge and best practices for collecting high-quality data and annotations for training machine learning models in agricultural contexts.
3. Develop proficiency in deploying deep learning models for real-world applications, specifically in the field of agriculture.
4. Experience collaborating with a diverse team to build an end-to-end applied AI solution.