Convolutional Free-space Optical Neural Networks for Image Recognition

With its unique parallel processing capability, optical neural network has shown low-power consumption in image recognition and speech processing. At present, the manufacturing technology of programmable photonic chip is not mature, and the realization of optical neural network in free-space is still a hot spot of photonic AI. In this letter, based on MNIST datasets and 4f system, one- and three-layer optical neural networks are constructed, whose recognition accuracy can reach 86.06% and 93.66%, respectively. Our network is better than the existing free-space optical neural network in terms of spatial complexity, and the three-layer's accuracy.