Microsoft supports Indian projects using AI to address ecological problems

Facial recognition is now fairly common in the consumer technology space. However, researchers from Delhi’s Indraprastha Institute of Information Technology (IIIT) are trying to develop a new sort of facial recognition technology: They are training computers to identify the faces of monkeys. No, not to help a monkey unlock a smartphone — but to aid civic workers in the national capital region to check the population of the primates.

Today, technologies such as machine learning and artificial intelligence (AI) are being used to manage environmental and ecological challenges. Thanks to the push by global tech giants such as Microsoft, these technologies are being applied to contain air pollution, preserve ecological balance, protect wildlife and so on.

Delhi has been grappling with the proliferation of monkeys caused by an abundance of food and the lack of any natural predators, leading to increased human-animal conflicts. While there are sterilisation programmes in place, these are often understaffed and underfunded. Moreover, civic workers have to capture the primates and transport them to a sterilisation centre to determine whether they have been sterilised already. This is a waste of resources as the same monkeys are often caught multiple times.

Ankita Shukla and Gullal Singh Cheema, two researchers from IIIT-Delhi, have taken up this problem as a challenge and are building an app powered by deep learning and AI to facilitate the identification of monkeys. The app will allow civic workers to identify sterilised primates by just clicking their photos on a phone. 

“We are using deep learning for this and training the model from the data we have. Our existing architecture has been modified for the monkey face and it seems to be working well right now with an accuracy rate of 98 per cent,”  says Shukla, who is pursuing PhD at IIIT Delhi. “Currently, we have only about 100 images of monkeys, but we’re collecting more data so that we can scale it up to 1,000. Then we shall have a decent estimate of how accurate it can be,” she adds. 

Shukla and Cheema’s work is one of seven cutting-edge projects in ecology and environmental conservation that has caught the eye of Microsoft’s  ‘AI for Earth’ initiative. The $50 million initiative started almost a year ago when Lucas Joppa, a PhD in Ecology, became the first chief environmental officer at Microsoft.

“We are democratising access to AI for everybody working in the environment space. If you want to use a machine learning approach in the conservation of water, climate or biodiversity, we want to hear from you,” says Joppa who spent eight years as a researcher at Microsoft before taking up this role.

The two biggest challenges that researchers like Shukla and Cheema face is the lack of data that can be tailored to their requirements. At present, the team is collaborating with the Wildlife Institute of India to collect  images of monkeys that can be used to develop the face recognition system. And they’re relying on Microsoft for the infrastructure, specifically the Graphic Processing Units (GPUs) which are required to run high-end computer vision algorithms.

EQWATER is another project which is supported by Microsoft in India. Headed by Mohan Kumar, a professor in the Department of Civil Engineering at Indian Institute of Science (IISc), Bengaluru, the project, which plans to use data modelling to ensure equitable distribution of water in cities, is part of a joint initiative by the IITs and IISc to solve science and engineering problems.

Starting off in South Bengaluru, an area that’s home to 1.5 million people, the researchers at IISc are working on developing a model to ensure all residents receive the same amount of water. Yogesh Simmhan, assistant professor at IISc's Department of Computational and Data Sciences and one of the researchers on the project, reveals that the work is being done in close collaboration with the Bangalore Water Supply and Sewerage Board (BWSSB).

BWSSB’s smart meters in pipelines across the city have given researchers the crucial data needed to build their models. This, when tallied with the actual consumption data from homes, gives them a picture of the shortages and over-supply of water in South Bengaluru.

“First, we want to understand the disparity in water distribution. Once we have figured out where the imbalances are, we’d like to see how we can rectify these issues,” says Simmhan.

The computational model that he and his team are building will be smart enough to detect where pumping additional water might not be possible due to old or leaky pipes. And this information can be used by the public works to determine where the most critical upgrades are needed. 

AI for Earth offers not just infrastructure — it could also deploy its own researchers or technologies to aid ecology and conservation efforts. Joppa says that India is turning out to be a hotbed for projects that are at the intersection of ecology and technology, largely because of the scale of the country. 

Perhaps the day is not far when AI will also be harnessed to check air pollution in Indian cities.

Microsoft’s AI for Earth programme

• IIT Kharagpur has joined hands with the Technical University of Munich to design a low-cost IoT device to help farmers diagnose diseases through image analysis
• The International Crops Research Institute for Semi-Arid Tropics is using AI to enhance pest forecasting
• Ashoka Trust for Research in Ecology and Environment is developing an AI-backed tool for documenting biological resources in North-East India