Artificial Intelligence for Food Image Recognition
Artificial Intelligence for Food Image Recognition
Today in India, diet-related Non-Communicable Diseases (NCDs) such as diabetes, cardiovascular diseases including coronary artery disease, and stroke are the leading causes of death, and over the past decade, the deaths attributed to NCDs have seen the greatest increase. At the same time, the latest report from India’s National Family Health Survey (NFHS), conducted between 2019 and 2021, shows that limited progress has been made in tackling undernourishment in the country, especially stunting and anemia.
Unhealthy diets underlie all these worrying trends for the diet-related NCD burden in India. However, there is still a lack of nationally representative data on dietary patterns and nutrient intake, the primary reason being the high cost and time associated with conventional dietary intake surveys. Timely diet data with appropriate geographic resolution is the need of the hour.
Anuvaad has proposed using a new app-based tool with AI-driven image recognition to convert pictures of food into specific, timely, high geographic-resolution dietary intake data. These data would then be mapped to enable anyone in the world to see what kinds of foods are eaten across India.
As a first step towards the development of an app-based tool for AI-based food image recognition, a comprehensive Indian food database is required. Currently, the Indian Council of Medical Research- National Institute of Nutrition (ICMR-NIN) Indian Food Composition Table (IFCT-2017) has data on around 600 raw foods mapped to their nutritional values, which is not sufficient to capture the huge and growing market of packaged foods, take-aways and the diverse range of cooked Indian recipes. Therefore, Anuvaad has worked on creating a comprehensive Indian food database, the ‘Indian Nutrient Databank’ (INDB), consisting of both raw food items and standard Indian recipes (around 1000) mapped to their nutritional values.
Anuvaad has collaborated with My Food Repo, a Swiss group for image/text recognition of Indian foods using Artificial Intelligence. The app uses either the barcode or a photograph taken by the user, coupled with an image recognition algorithm, to identify what the consumer is eating.
Through this project, Anuvaad aims to use a Citizen Science approach to crowd-source diet data to fill the huge gap of diet data existing in India and thus build a diet-data map of India that demonstrates the huge diversity of foods consumed across India at a fine resolution.
Artificial Intelligence for Food Image Recognition
Today in India, diet-related Non-Communicable Diseases (NCDs) such as diabetes, cardiovascular diseases including coronary artery disease, and stroke are the leading causes of death, and over the past decade, the deaths attributed to NCDs have seen the greatest increase. At the same time, the latest report from India’s National Family Health Survey (NFHS), conducted between 2019 and 2021, shows that limited progress has been made in tackling undernourishment in the country, especially stunting and anemia.
Unhealthy diets underlie all these worrying trends for the diet-related NCD burden in India. However, there is still a lack of nationally representative data on dietary patterns and nutrient intake, the primary reason being the high cost and time associated with conventional dietary intake surveys. Timely diet data with appropriate geographic resolution is the need of the hour.
Anuvaad Solutions is a registered LLP (ID Number AAY-4683). All project initiatives are supported by the Bill and Melinda Gates Foundation.
Follow Us
Contact Us
Anuvaad Solutions LLP
Office No-1543, Regus,15th Floor,Eros Corporate Tower, Nehru place,New Delhi 110019.
Anuvaad Solutions is a registered LLP (ID Number AAY-4683). All project initiatives are supported by the Bill and Melinda Gates Foundation.