Simple wrapper of OpenCV‘s deep neural network implementation.

To process an image requires three steps:

  1. Initialise NeuralNetwork object, supplying configuration file, weights file, computation backend and a computation target (latter two default to CUDA if no specified)
  2. Call process method with a valid image
  3. 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


View Github