How to modify OpenTLD to support object recognition?

[tianshiz]

From what I have gathered, OpenTLD is a tracker and detector, it tracks what you tell it to and detects the object frame by frame. I feel that this can be applied to the broader scope of object recognition, where the software learns many object and can name the objects it sees in a frame. I'm aware of openTLD's one object tracking limitations, but there shouldn't be a limitation to find all the recognizable objects correct? It should be able to go through the database of saved objects and look for those objects in a frame.

Any thoughts on this? Can this be done easily? Better yet, has anyone made modifications for this already?

thanks

[hujilin1229]

hello, I have the same opinion with you and can we discuss about it together?