Real Time Human Detection and Counting with OpenCV and YOLOV3. The model used for detection is YOLOV3. All weights can be downloaded here: https://pjreddie.com/darknet/yolo/ or from this repo: yolov3.weights -> Download file (254 MB).
For accurate results, use a video...
- OpenCV-python
- Numpy
- imutils
- Argparse
- Sklearn
Use git clone to clone the repo and other pre-requisites. Install and Add to PATH OpenCV (with contrib modules) together with other pre-requisites.
- Crowd Density and event count over a period of 2 hours
- Crowd Distribution Free Movement around a museum
- Crowd Distribution during an Robotic Development Competition
Output will be displayed in a new window created by OpenCV, but can also be viewed in the videos above or inside the Output directory.
Current limitations of the detection algorithm can be found inside the custom_yolo_model directory.