network.h - justinmgarrigus/VGG-Simulation GitHub Wiki

Properties

int layer_count: How many layers are in this network.

layer **layers: The list of layers to be operated on. Should be of length layer_count.

char **labels: String representations of the output layer. This assumes the output is a one-dimensional vector.

Functions

network* network_create(char* data_file, char* label_file): Creates a network given the .nn file that stores the pre-trained weights and the .json file that stores the labels.

void network_free(network* network): Frees the data stored in the neural network, including each of the network's layers.

void network_feedforward(network* network, void* inputs): Performs the feedforward operation on the network, inputting the data into the first layer.

void network_decode_output(network* network, int count, char** predictions): To be invoked after a feedforward operation; places the top count predictions into the predictions array.