avnishgupta143 / multithreaded-object-detection-and-tracking-cpp Goto Github PK

Contains implementation of multi-threaded object detection followed by tracking

CMake 2.26% C++ 97.74%

yolo3 yolo deep-learning multiple-object-tracking detection kalman-filter hungarian-algorithm hungarian multiple-object-detection multithreading

multithreaded-object-detection-and-tracking-cpp's People

Stargazers

Watchers

Forkers

zhangyuelong amk2777 temi92 shubhamvad computervisionworks irwanmazlin zhouleidcc vck chadyuu swcho33 barbecacov windwing2016

multithreaded-object-detection-and-tracking-cpp's Issues

Running @ 1477msec /frame on Nvidia Jetson TX2 - Isn't it too slow?

Hi, I am trying to benchmark the system on Nvidia Jetson TX2 with weights and cfg files downloaded from original authors website as mentioned in your instructions and image size of 608x608.

It seems to be taking approx. 1.7 seconds per frame for Detection+Tracking! Isn't it a bit too much time as the standard Yolov3 implementation roughly takes approx. 3.7 FPS on the same hardware.

Would you like to suggest some change in setup so that the execution performance could be improved?

Recommend Projects