The debate what is biryani, and which is the best one remains unsettled among the gourmands in India. But, a clutch of researchers in Hyderabad have used AI tools to analyse video recordings of the cooking processes to conclude that there is a systematic and detectable difference among the variations.
The researchers at International Institute of Information Technology, Hyderabad used visual learning models to analyse and reach the conclusion about the variations.
The dataset used by the researchers included 12 distinct types of biryanis: Ambur, Bombay, Dindigul, Donne, Hyderabadi, Kashmiri, Kolkata, Awadhi, Malabar, Mughlai, Sindhi, and Thalassery. It does not include the Moradabadi variant that uses a distinct shape of the cauldron.
The researchers tried to understand what makes the biryanis unique and how: “By comparing the cooking process for different types of biryani, we can identify common patterns and variations in the cooking methods, ingredients, and techniques used. This can help us understand the unique characteristics of each biryani recipe and how they differ.”
The paper titled ‘How Does India Cook Biryani’ lists its authors as C.V. Rishi, Farzana S., Shubham Goel, Aditya Arun and C.V. Jawahar, who is a faculty member at the IIIT-H, was presented at the ‘Indian Conference on Computer Vision, Graphics, and Image Processing’ at Mandi in December 2025.
Despite the variations in preparation processes, the researchers identified some common steps: “canonical procedural frames extracted from videos, including soaking rice, marinating chicken, frying onions, and adding spices—steps that recur across multiple biryani variants despite regional differences.”
While the research is focused on visual learning models, the research may prove useful for Geographical Indication tagging that has bypassed the popular culinary dish in different parts of the country. The GI tag bid for Hyderabadi Biryani was abandoned after the Deccan Biryani Makers Association refused to share the cooking process that is key to understanding as well as securing the coveted branding.
Curated by Shiv Shakti Mishra
