Skip to content
Sergey Nuzhny edited this page Aug 10, 2023 · 4 revisions

The base examples (you can change parameters and use advanced options):

  1. Mouse tracking example:

MultitargetTracker.exe -e=0 -o=res_mouse.mp4

  1. Motion detector with background/foreground segmentation:

MultitargetTracker.exe ../../data/atrium.avi -e=1 -a=0 -o=res_bgfg.mp4

  1. Haar face detection and tracking:

MultitargetTracker.exe ../../data/smuglanka.mp4 -e=2 -a=0 -o=res_face.mp4

  1. HOG pedestrian detection and tracking:

MultitargetTracker.exe ../../data/atrium.avi -e=3 -a=0 -o=res_pedestrians.mp4

  1. Base detector with opencv_dnn backend. MobileNetSSD weights as default but any other compatible NN can be used. You must download manually and set paths to cfg and weight file in examples.h in OpenCVDNNExample::InitDetector method:

MultitargetTracker.exe ../../data/atrium.avi -e=4 -a=0 -o=res_ocvdnn_mobilenet.mp4

  1. YOLO detector with Darknet backend (build project with -DBUILD_YOLO_LIB=ON). YOLO weights you must download manually and set paths to cfg and weight file in examples.h in YoloDarknetExample::InitDetector method:

MultitargetTracker.exe ../../data/atrium.avi -e=6 -a=0 -o=res_darknet_yolo.mp4

  1. YOLO detector with TensorRT backend (build project with -DBUILD_YOLO_TENSORRT=ON). YOLO weights you must download manually and set paths to cfg and weight file in examples.h in YoloTensorRTExample::InitDetector method:

MultitargetTracker.exe ../../data/atrium.avi -e=7 -a=0 -o=res_tensorrt_yolo.mp4

  1. CarsCounting example works with:
  • YOLO network and TensorRT or Darknet backend (build project with -DBUILD_CARS_COUNTING=ON -DBUILD_YOLO_LIB=ON -DBUILD_YOLO_TENSORRT=ON)
  • any NN and opencv_dnn backend YOLO weights you must download manually and set paths to cfg and weight file in settings.ini file. Lines configuration is in geo_bind.ini

MultitargetTracker.exe video_with_traffic.mp4 -e=7 --geo_bind=map/geo_bind.ini --settings=settings.ini -o=cars_count_speed.avi

  1. AsyncDetector also works with YOLO network and Darknet backend (build project with -DBUILD_ASYNC_DETECTOR=ON -DBUILD_YOLO_LIB=ON). YOLO weights you must download manually and set paths to cfg and weight file in AsyncDetector.cpp in AsyncDetector::CaptureThread method:

AsyncDetector.exe ../../data/atrium.avi -g=0

Clone this wiki locally