Street Monitor

Keeps track of pedestrians and vehicles in the road.
Tracks the movement and direction of all elements in the road.
Estimates the speed of each object and a direction vector.
Tracking performed based on object centroid position obtained from background subtraction and blob segmentation.

Download OpenCV 4.2.0 and install it on your system (use scripts/install-opencv.sh).
- For optimal performance ensure that optimizations are enabled.
- If a NVIDIA is used ensure that the CUDA SDK is installed.
Install CMAKE and setup your IDE for C++ development.
Install GCC for linux development or Visual Studio for windows development.
Dependencies can also be obtained from the conan package manager (https://conan.io/center/)
- To install dependencies run conan install ..

Data for testing can be downloaded from youtube.
The file scripts/dataset.sh can be used to obtain test data.
Test video data is obtained from youtube might be suject to licensing and not available for commercial usage.

YOLO model is used to classify moving objects such as cars and pedestrians.
YOLO V5 is available on https://pytorch.org/hub/ultralytics_yolov5/ / https://github.com/ultralytics/yolov5, check the latest releases on github.
- The PyTorch models have to be converted into ONNX files.

This project is distributed under MIT license and can be freely used for commercial applications.
Project license is available on the Github page.

tentone/street-monitor