Abstract:In the face of largescale image recognition tasks, the deep learning method based on convolutional neural network has a problem of too long training time, resulting in low recognition efficiency. Therefore, a largescale image efficient recognition algorithm based on local feature depth belief network is proposed. First, a plurality of local features are extracted from the original image, and each local feature is classified according to the label assigned to the image. Then, the deep belief network is trained by using the local features of the classified image to obtain the relevant parameters of the network. Finally, the deep belief network is used for image recognition. Image recognition experiments were carried out on the CASPEALR1 largescale image dataset. The experimental results show that the proposed algorithm, which has good accuracy and efficiency, is superior to other deep learning methods.