Ballistics Analysis with Deep Learning

During my MSc studies at the University of Alberta, I developed a tool for ballistic analysis, enhancing the traditional manual approach with deep learning and image processing techniques.
The conventional method for ballistic analysis involved manually locating the bullet in each frame and estimating its displacement over time using a visual background reference. To improve accuracy and efficiency, I implemented a deep learning-based image detection system to automatically track the bullet’s position. Using image processing techniques, my tool was able to calculate:
-Bullet velocity
-Trajectory
-Yaw and pitch angles

This project allowed me to apply research methodologies, work with deep learning models, and collaborate outside my research group. It was a valuable experience in computer vision, automation, and data analysis.

More details & source code: https://github.com/reyesyan/Ballistic_Experiment_Analysis