With small changes to the vision class, it would be possible to support multiple algorithm for vision processing and feature extraction.
The user could then choose the desired implementation at runtime, via the menu.
examples of algorithms could be:
Original (CPU)
Original (GPU)
Bruteforce
I think that allowing multiple of these algorithms to be implemented side by side, will increase the speed of testing new algorithms.
With small changes to the
vision
class, it would be possible to support multiple algorithm for vision processing and feature extraction. The user could then choose the desired implementation at runtime, via the menu.examples of algorithms could be:
I think that allowing multiple of these algorithms to be implemented side by side, will increase the speed of testing new algorithms.