During my summer break in 2022, I interned at Nullspace Robotics, a robotics and programming enrichment centre for young students. As part of the software required for an automated lego pieces sorter project, I wrote Python scripts to capture and crop images using object motion detection and bounding box image processing.
Afterwards, I built and trained image classification CNN model using Tensorflow, achieving 93% testing accuracy.
Finally, by integrating the Tensorflow model with a live feed from a RaspberryPi camera, it could classify lego pieces in realtime to a high degree of accuracy. You can check out the source code and preview in the link below!
Platform

Stack






