AgriVision AI leverages the power of Artificial Intelligence (AI) and Machine Learning (ML) to assess the impact of floods on agricultural productivity. Our project specifically addresses the devastating floods that occurred in Kerala, India in August 2018, and aims to provide actionable insights for improving disaster preparedness, response strategies, and agricultural planning.
On 16 August 2018, severe floods affected the south Indian state of Kerala due to unusually heavy rainfall during the monsoon season. It was the worst flood in Kerala in nearly a century, leading to it being declared a “Level 3 Calamity” by the government. Nearly half of Kerala's population depends on the agriculture sector for livelihood. The floods caused significant damage to human lives, property, flora, fauna, agriculture, economy, and food security.
- Economic Losses: The plantation industry was at risk of losing up to EUR 88 million and approximately 40% of the current crops.
- Crop Damage: Rice paddy cultivation suffered significant damage, with 26,106 hectares of farmland affected.
- Other Crops: Tea, rubber, cardamom, and black pepper also experienced adverse effects. In specific regions like Nilambur, Malappuram, and Kalikavu districts, an estimated 500 acres of plantation land were destroyed due to landslides.
- Food Production: The damage disrupted food production, impacting not only Kerala but also having a ripple effect throughout India.
- Human Development: Rapid human development, encroachment on rivers, and environmental degradation exacerbated the flooding.
- Wetland Encroachment: Construction on wetlands reduced their natural ability to store water, leading to poor surface runoff management.
- Construction Surge: Increased construction activities, including houses, buildings, and tourist resorts, contributed to the severity of the flooding.
AgriVision AI aims to assess the impact of these floods on agricultural productivity using Normalized Difference Vegetation Index (NDVI) data. By analyzing NDVI data from May to September over several years, we aim to provide insights into the extent of agricultural damage and recovery post-flood. The results will help local communities and policymakers improve disaster preparedness, response strategies, and agricultural planning.
We utilized Azure Machine Learning Studio provided by Microsoft for building this project due to the following reasons:
- Foundational Models: Using pre-trained models and other state-of-the-art models.
- Speed & Efficiency: Saves cost to build from scratch and speeds up the process.
- Customization: Feature to customize/fine-tune the model using scripts, and more.
You can run the notebook to see out analysis and make your own edits if you desire. Access the notebook through this link. You'll want to download it and upload it to your Azure ML Studio: https://drive.google.com/file/d/1wVRXX2x9PBD1KzhaBceRgO55wQC_B2MB/view?usp=sharing
This project is licensed under the MIT License - see the LICENSE file for details.