Using YOLOV3 to detect, classify, and label the objects.
- At first, we need to load the cnn.
- Convert each frame read from camera into blob type because cnn can only handle blob type images.
- Set the converted blob type image as an input to the network.
- Now that we have provided the input we require an output.
- There are three outputs for a single detected objects.
- Find the name of the output layers to find the outputs of the three layers.
- Compute bounding box coordinates, class name, and confidence fo the detected object.
- Draw rectangle, put label and confidence over the detected object on the frame.
Look at the manual provided in this repo for detail explanation.