Tech Stack
Python
PyTorch
OpenCV
Image Processing
FastAPI
Machine Learning
Description
This is not meant to be a toy food-classification project. The larger goal is to build a modular and production-minded food analysis pipeline that can move from food recognition toward portion estimation and more realistic nutritional analysis.
The project is being designed as a scalable CV system with separate stages for segmentation, classification, estimation, and nutrition lookup. A big focus is on making the architecture flexible enough to support multiple datasets, adapters, and future model upgrades without rewriting the whole pipeline.
- Designing a modular computer vision pipeline for food segmentation, classification, and nutrition estimation
- Planning dataset adapters and unified schemas so multiple food datasets can be added cleanly
- Working toward portion-aware analysis instead of just top-1 food classification
- Using PyTorch and OpenCV as the core stack for training and image-processing experiments
- Building the system in a way that is portfolio-worthy, extensible, and closer to production thinking
Page Info
Pipeline Architecture
High-level architecture showing the staged computer vision and nutrition analysis flow.
Training / Dataset Workflow
Dataset, adapter, and model-training workflow for segmentation and classification experiments.
