Convolutional neural networks provide a powerful solutionfor image recognition, but also present a trade-off due the computingpower required to carry out the classification procedure. This is primarilydue to the difficulty involved in calculating multiple convolution operations over various layers in the neural network. With the ever increasingperformance of mobile devices, it is natural to explore the possibility ofimplementing such a system on board a device to perform interactiveimage recognition tasks in real time.
© 2008-2024 Fundación Dialnet · Todos los derechos reservados