This repository contains a set of small, focused experiments exploring visual perception and spatial understanding using computer vision techniques.
The goal is not to optimize for benchmark performance, but to understand how different approaches behave under practical constraints such as lighting variation, motion, noise, and limited data.
Each experiment documents the problem setup, approach, observations, and trade-offs, with an emphasis on why certain methods work or fail in applied settings.
- Classical and learning-based computer vision methods
- Small-scale perception pipelines
- Emphasis on experimentation and observation
Work in progress. Experiments will be added incrementally.
MIT