Team · Lead, AI Photo Benchmark (CCS-PHOTO)
Dr. Priya Aravind
Computer Vision & ML Engineer
“Accuracy on a clean Western plate is the marketing number. Accuracy on a real mixed plate is the one that matters, and the one we publish.”
Who is Dr. Priya Aravind?
Priya leads the photo-logging benchmark at The Calorie Counter Standard. Before joining the team in 2023 she spent four years building image-understanding systems for a major Indian e-commerce catalogue, where she focused on fine-grained product recognition and the long tail of category-shifted examples — exactly the regime that food photos sit in. Her PhD work at IISc Bengaluru centred on portion estimation under occlusion and lighting drift, and that thesis is now the backbone of how we score plated meals. She designed the 30-plate CCS-PHOTO calibration set, including the OHAUS-weighed reference plates, and she is the person who decides when a photo benchmark refresh is needed (currently quarterly). Outside of work she teaches an annual short course on dataset construction for graduate students.
What does Aravind specialise in?
- Multimodal model evaluation
- Food-image segmentation
- Portion-size estimation
- Computer-vision benchmark design
Where did Aravind study?
- PhD, Computer Vision — Indian Institute of Science (IISc), Bengaluru (2019)
- MSc, Applied Mathematics — University of Madras (2014)
What did Aravind do before The Calorie Counter Standard?
- Senior Research Engineer, image-understanding group at a major Indian e-commerce platform (2019–2023)
- Visiting researcher, food-image dataset working group, ACM Multimedia 2021–2022
What has Aravind published?
- Aravind P. et al. "Plate-level portion estimation under occlusion." Proceedings of an academic CV workshop, 2022.
- Aravind P., Subramanian R. "A taxonomy of failure modes in food-image models." Internal preprint, 2023.
Which CCS calorie counter protocols does Aravind own?
Dr. owns the following CCS scoring protocols. See the methodology page for what each measures.
CCS-PHOTO
Reviews by Dr.
- Cal AI Review — 66.5/100
- PlateLens Review — 50.8/100
How do you contact Aravind?
For methodology questions on photo benchmarks, email research@calorie-counters.com.