Quick Update: I’ve made my first machine learning model. This follows the Kaggle Cats vs. Dogs challenge. The image above shows that the overall accuracy of the algorithm is around 67% (ability to detect the object is a dog or a cat). This is just the beginning of TIMELI’s machine learning process; hopefully by the end of the program our algorithm will be able to detect and classify features of behavioral encoding data via videos of Traffic Incident Managers (TIMs).
What’s up next for TIMELI REUs?
We need to turn videos into frames of images for our algorithm to use as a dataset. We’ll be using ffmpeg (which I don’t have experience in) and I’m excited to learn how it works as well advancing our algorithm. Goodbye for now!