Aim of this project is to exceed the current state-of-art systems in tracking objects such as signs which have limited in class variation. Right now I'm working on modified code in order to speed up generic object tracking. It is expected that we can speed up the system further using
- Higher power processors and GPUs
- Simplifying the neural nets using the knowledge of limited in-class variation.