shrutimalik123/the-pill-counter-computer-vision-object-count-engine

The Pill Counter is an educational Python simulation that teaches simple linear regression through a pretend pharmacy automation scenario. You play a robotics engineer building an overhead camera system that estimates how many pills are on a sorting tray based on the total pixel area a camera detects, rather than counting each pill directly. The project's real goal is teaching the math behind linear regression rather than building an actual vision system. It covers simple linear regression as a way to model a straight line relationship between one input value and one numeric output, ordinary least squares as the method used to find the best fitting line by minimizing the squared distance between predicted and actual points, and the idea of producing a continuous number as output instead of sorting things into fixed categories. It also touches on residual analysis, which looks at how far off each prediction is from the real recorded value. Gameplay walks through a calibration data set that pairs past pixel area measurements with hand counted pill totals, watches the script calculate the slope and intercept of the best fit line from that data, then feeds in a new pixel area reading from a live scan and produces a predicted pill count, rounded to a whole number since pills are physical objects. Like its companion project, this one is built with zero external dependencies, using only core Python and basic algebra with no matrix or machine learning libraries. To run it you need Python 3, then you clone the repository, save the script as pill_counter.py, and run it from the command line with no extra setup.