Tag Archives: classify

Green warriors from India receive Microsoft AI for Earth grants to enable a sustainable future – Microsoft News Center India

With a trained AI algorithm, the team hopes to classify the urban and rural areas, identify forest cover, river beds and other water bodies from satellite images, and create a precise grid map for the region. The team hopes to apply computer vision to create a comprehensive database of biodiversity in the region to help policymakers and local communities make better-informed economic, ecological, and infrastructure-related decisions.

“You can’t save an ecosystem if you don’t fully understand it,” exclaims Dr. Mariappan. “That’s where our data along with Microsoft’s AI resources can help.”

Tracking the monkey population in urban areas using AI-powered image recognition

A woman sitting on a table with a coffee cupThe monkey population in urban India has spiraled out of control in recent years. India’s capital city, New Delhi, alone reports at least five cases of monkey bites daily that can cause rabies and be fatal. It is estimated that 7,000 monkeys prowl the streets of the capital, damaging public property and attacking people. With their natural habitat shrinking owing to urbanization, authorities are struggling to avoid monkey attacks.

Managing the growth of the population is critical. Currently, there is no way to identify which monkeys have already been given birth control or sterilized without further handling such as tattooing a code or embedding a microchip in the monkeys. Ankita Shukla, a PhD student at Indraprastha Institute of Information Technology Delhi (IIIT Delhi), aims to use computer vision as a non-invasive alternative for identifying and tracking monkeys as it is safer and less stressful for the animals, as well as humans.

Shukla, a native of a small town near Lucknow, had earlier worked with the Wildlife Institute of India on a project to classify endangered tigers in a nature reserve with machine learning and distance-object recognition algorithms. She wants to combine this experience in wildlife monitoring with machine learning to create a tangible solution for the simian problem in cities.

She is creating an AI-enabled app that can help the community tag monkeys in photographs and upload it to a cloud where authorities can track the simian population’s growth, vaccination history, and movements. “With a bird’s eye view of the monkey population, we can deploy contraceptives more efficiently,” she says. “Training a deep neural network with image recognition to identify a monkey and its species, and whether it’s already been sterilized could go a long way towards solving this crisis,” Shukla adds.

Having teamed up with Saket Anand, a professor at IIIT Delhi, she pitched the idea to the AI for Earth panel earlier this year. The team plans to leverage the Microsoft Azure platform for the processing power required to train the AI model.

“The Microsoft resources and technical assistance helped us develop a genuinely useful app,” says Shukla. “We’re now trying to take things to the next level so that we can find a solution to the monkey menace in a scientific and humane manner.”

Office 365 labels help keep content under control

Office 365 labels make it easy to classify data for compliance purposes through both manual and automatic methods.

Office 365 label policies, included with E3 subscriptions, provide a central location to configure and publish labels to Exchange Online, SharePoint Online and the services that depend on them, such as Office 365 Groups.

For example, administrators can add a label named Financial Data to the Security & Compliance Center and designate it to keep data with that label for six years. An Office 365 label policy pushes that label out to the other Microsoft services on the platform.

Users mark items in their inbox or documents with labels. Office 365 labels have policies that retain or delete data based on the organization’s needs. Personal data might get marked for deletion after a certain amount of time following a review, for example, or other information might get marked as an organizational record, so nobody can change or purge it.

The Advanced Data Governance functionality in an E5 subscription enables the automatic application of data labels based on keywords or sensitive information. Policies could mark all data with Social Security numbers as personal data or mark all data with credit card numbers as financial data.

The Advanced Data Governance functionality in an E5 subscription enables the automatic application of data labels based on keywords or sensitive information.

Label policies require some forethought to cover different types of information. Many organizations might require multiple labels to cover the types of data to retain or delete.

Office 365 labels take approximately 24 hours before they appear. Automatic labeling starts after about seven days.

User or organization-wide retention policies that hold data take precedence over Office 365 labels. A policy that holds data for 10 years across the organization will overrule one that removes certain data after five years.

View All Videos