Simple wrapper of OpenCV's deep neural network implementation.
To process an image requires three steps:
- Initialise NeuralNetwork object, supplying configuration file, weights file, computation backend and a computation target (latter two default to CUDA if no specified)
- Call
process
method with a valid image - Call
detections
method to then retrieve a list of struct containing what object and where it was detected and the confidence of said detection.
See LICENSE.md for licensing details