Open 3073 opened 6 years ago
The speed depends upon the detector and tracker. if the detector is initialized once in few frames and the tracker is running for the rest of the frames, then it may be quicker. What is the application you are applying? Is it any video that you are processing or screen grabbing?
it is a video. but meanwhile i will apply it for real time (with raspberry pi camera) the problem with initializing the tracker with the detector once do not allow you to detect a new object in the next frame. for instance if you initialize the tracker with the bounding boxes of the detector in the first frame and then allow the detector to take rest ( do not detect) then you will not have the chance to detect a new object that appear in the third or fourth frame.... I tried both... but the speed is the matter....I will share you how it is working...
Normally i am using this machine and I think it is enough.
On Sat, May 19, 2018 at 9:39 PM, Vishnu Pradeesh notifications@github.com wrote:
The speed depends upon the detector and tracker. if the detector is initialized once in few frames and the tracker is running for the rest of the frames, then it may be quicker. What is the application you are applying? Is it any video that you are processing or screen grabbing?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pradeeshenator/Computer-Vision/issues/1#issuecomment-390457306, or mute the thread https://github.com/notifications/unsubscribe-auth/AgXgdfD8jcuV4nCnTG5WA6CGFJxYsQ0zks5t0POAgaJpZM4UF89a .
OH sorry .... Application: I am using it for passenger counting
thank you for that....I used SSD detector to initialize and dlib tracker and it works but it is too slow... how can make it fast? please help with any kind of pointer..