IIT-Bombay has developed SpADANet, an AI model that rapidly assesses cyclone and disaster damage using drone and satellite images. Designed to work on low-compute devices, the model helps disaster agencies quickly categorise damage and improve response, even in resource-limited regions.

Mumbai, Dec 20: IIT-Bombay has released an AI model to assess the damage caused by cyclones and other calamities with accuracy and speed. The spatially aware model analyses images from drones and satellites to categorise damage from minor impact to destroyed infrastructure, which would assist disaster management teams to take timely action in their response.

Heavy monsoon damage highlights need for faster assessment

This monsoon, several episodes of heavy rainfall wreaked havoc in pockets of Maharashtra. The infrastructure, agricultural and other destruction in such events is assessed using drones and other aerial methods. The assessment of these multiple visual data sets is difficult and time-consuming for the human eye.

SpADANet analyses damage across locations

The spatially aware domain adaptation network (SpADANet), the new AI framework developed by IIT-Bombay, assesses all kinds of damage accurately at different locations, which helps disaster management authorities to better manage calamities and their after-effects.

Self-supervised learning strengthens model accuracy

“SpADANet first teaches itself by studying unlabelled images from a domain (hurricane study area) by employing a process called self-supervised learning. This helps the model understand general visual patterns, such as how undamaged and damaged buildings or debris appear in aerial photos. By the time it sees labelled data, it already has a strong sense of what to look for in the data,” elaborated Prof. Surya Durbha, who led the study.

Images tagged into four damage categories

The SpADANet model quickly tagged images into the categories of no damage, minor damage, major damage or destruction, which aided a quick response and action plan. The model does not just look at colour and shape. As it is spatially aware, it recognises damage patterns by location, wind patterns and context.

Optimised for use on low computing devices

Further, the researchers have optimised the model to run with limited computing power, and it can be used on tablets and phones, making it a handy tool in the field and addressing a genuine bottleneck in disaster response, particularly for regions with limited resources.

Model helps overcome key disaster response challenges

“Agencies like NDMA (National Disaster Management Authority in India) face three main constraints: lack of labelled data, limited computing resources, and regional differences (domain gap) in image characteristics. SpADANet can help overcome these barriers as it learns from fewer labels, adapts to new regions, and can run on modest hardware once trained. With continued collaboration between researchers and government agencies, such AI models can soon become part of near-real-time disaster response systems,” said Pratyush Talreja, a PhD candidate and first author of the study.

Also Watch:

Earlier IIT-Bombay model aided Mumbai flood preparedness

Earlier this year, the institution’s local weather prediction model, mumbaiflood.in, equipped Mumbai’s civic body to prepare for waterlogging in the city.

To get details on exclusive and budget-friendly property deals in Mumbai & surrounding regions, do visit: https://budgetproperties.in/