Generate Yolo2 config for custom dataset.

bind variable values.

Line 7: classes=20, the number of categories we want to detect

Line 8: batch = 64, this means we will be using 64 images for every training step

Line 9: subdivisions = 8, he batch will be divided by 8 to decrease GPU VRAM requirements. The training step will throw a CUDA out of memory error so you can adjust accordingly.

Run python Yolov2Generate.py

C ++ A wrapper for the Darknet detector.

Install OpenCV.

Create a YoloV2 object with parameters thresh and hier_thresh. Parameters thresh and hier_thresh thresholds of probability.

Load the network by passing the configuration file addresses to the parameters by calling method loadNet(const std::string & model_prototxt, const std::string & model_net, const std::string & model_class).

To process images, call the imgProcess (cv :: Mat img) method.