Template Matching
Lee Cognition and Memory Lab
January 2025 — May 2025
Overview
Worked as a Research Assistant to automate the analysis of behavioral data from screen recordings. The project involved high-precision tracking of user interactions to support cognitive research.
Strategies Used
- OpenCV: First began by utilizing OpenCV template matching to locate cursor positions within 100ms intervals of screen recordings. From trial and error, we realized that the accuracy OpenCV had was not enough for our use case.
- Binarization: An idea we ran into was to turn the frames into black and white pixels based on a brightness threshold before beginning analysis. This helped increase the accuracy and allowed us to use direct template matching.
- Multithreading: This technique was used to help speed up the process of running the program. Simultaneously while processing frames, the frames that have been processed would be analyzed, significantly reducing run time.