In charge of developing a hardware and software road/runway surface quality inspection system for detecting potholes and imperfections using depth cameras and image processing.
I worked with Intel D and L series depth and lidar cameras and a Jetson Nano dev kit.
We created our own dataset by collecting raw data from the streets and running it through a 10-stage image processing pipeline, that I designed and implemented, to produce bounding boxes of potholes and stitched sections of road.
The system containing the Jetson Nano and the cameras was mounted on a push-cart that I built.
My team consisted of 2 more software developers, one of whom worked on a React front-end application, and the other one worked on integrating a Flask backend with the pipeline that I had developed.

Key learning points that I developed:
- stereo depth processing and image acquisition
- 3D plane fitting (by linear regression)
- image filtering (smoothing, thresholding, morphological operations)
- sparse optical flow (Lucas-Kanade method)
- image stitching (and cropping)

I gained experience with the following tools and technologies:
- Python 3, OpenCV, Intel RealSense SDK, NumPy, PyCharm IDE, Git

Hardware devices used:
- Intel RealSense D435 and L415 Stereo and LiDAR cameras
- nVidia JetsonNano Development Kit running Ubuntu Linux and Intel RealSense SDK for interfacing the cameras
- Personal development machine running Windows

Part of the Mentorship Program "Discover your passion in IT".
During a 7 month period I learned the basics of Android app development.
Together with my team, we implemented an app called "BusTracker", designed to find you the quickest bus route between two points in the city of Cluj-Napoca, using the routes of the public transportation system.