A machine learning algorithm developed by an innovative Indian Forest Service officer is set to redefine forest management in Madhya Pradesh. This pioneering approach, powered by satellite imagery, aims to curb deforestation through an Artificial Intelligence (AI) alert system.
The Visionary Behind the Innovation
Akshaya Rathore, the Divisional Forest Officer (DFO) of Guna, has spearheaded this transformative project. An alumnus of IIT Roorkee, Rathore combined his technical acumen with the capabilities of Chat GPT to create the Real-Time Forest Alert System. This system is being launched as a pilot project in the districts of Shivpuri, Guna, Vidisha, Burhanpur, and Khandwa—areas notorious for illegal felling and encroachment.
How the System Works
The cloud-based system integrates satellite data, machine learning algorithms, and real-time field feedback to monitor forest areas with unmatched precision. Capable of detecting changes in areas as small as 10×10 meters, the system provides instant alerts to DFOs when unauthorized activities, such as crop cultivation, construction, or land use changes, are detected in forested regions. These alerts prompt immediate instructions for ground verification, ensuring swift action against illegal activities.
A First for Madhya Pradesh
This initiative marks the first instance of AI being employed to tackle deforestation in the state. According to a 2023 report by the Forest Survey of India, Madhya Pradesh boasts the largest forest and tree cover at 85,724 sq km, but it also reported the highest loss of forest land, amounting to approximately 612.41 sq km.
Inspiration and Development
Rathore’s journey began after witnessing the tangible effects of climate change. “My research on AI, including published papers, inspired this project,” he stated. A significant impetus came from addressing the Guna encroachment issue, where he validated a paper-based model through data confirmation.
The catalyst for Rathore’s AI model was a land dispute between two communities during Diwali last year, which tragically resulted in the death of a community leader and subsequent arson. “This tragedy highlighted the urgency of developing an AI-based solution for proactive land management,” Rathore noted.
Technological Advancements
Initially, forest monitoring relied on manual methods and ground-level confirmation. To enhance accuracy, Rathore introduced a dynamic AI model, writing the basic alert generation code using Python. While ChatGPT assisted with scripting, he designed the core architecture himself. “Python required specific adaptations for tablet-based deployment,” he explained.
Rathore’s mechanism was further refined by studying a similar alert-based system in Karnataka. “Our AI system aims to generate alerts every 2-3 days, significantly improving over Karnataka’s 21-day alert frequency model,” he stated. “Before the monsoon, we plan to establish a comprehensive database to fully train the model.”
Once the system achieves a 99 percent success rate, it will function as a predictive tool for land and resource management. This includes identifying illegal activities such as felling, encroachment, and optimizing budget allocation and manpower deployment.
Future Prospects
The current system ensures staff accountability by requiring photo uploads from within alert zones. However, full automation in Phase 2 will reduce human dependency. “We hope to introduce drones that will take over human tasks,” Rathore shared. “The database must be expanded to include year-wise data for robust time-series modeling, minimizing errors from seasonal variations.”
The deployment of this AI system signifies a significant stride towards preserving Madhya Pradesh’s rich forest cover, leveraging technology to safeguard natural resources for future generations.
Note: This article is inspired by content from https://indianexpress.com/article/india/how-a-madhya-pradesh-ifs-officer-is-fighting-deforestation-with-some-help-from-ai-tool-he-developed-9973558/. It has been rephrased for originality. Images are credited to the original source.