Image target recognition method based on optimized convolution architecture

Abstract

The invention discloses an image target recognition method based on optimized convolution architecture. The image target recognition method includes collecting and enhancing an input image to form a sample; training the sample on the basis of the optimized convolution architecture; performing classified recognition on an image target by using the convolution architecture after training, wherein optimization of convolution architecture includes ReLU activation function; locally responding to normalization; overlapping and merging a convolution area; adopting neuron connection Drop-out technology; performing heuristic learning. Compared with the prior art, the image target recognition method has the advantages that tape label samples can be expanded, and the image target recognition method is supportive of classification of many objects and acquiring of high training convergence speed and high image target recognition rate and has higher robustness.

Description

technical field [0001] The invention belongs to the technical field of image analysis and recognition, and in particular relates to an image target recognition method based on an optimized convolution architecture. Background technique [0002] At present, pattern recognition, as a hot research field in machine learning, has been paid more and more attention with the massive appearance of computer network image data. In order to effectively manage image data and serve users better, automatic recognition of image objects becomes particularly important. [0003] Traditional image recognition methods such as SVM (Support Vector Machine, Support Vector Machine) and Boosting mainly perform classification and recognition by extracting image features. The convolutional neural network is a deep learning model based on supervision. Its essence is to simulate the human brain mechanism to build a learning network with multiple hidden layers. Transformations such as translation, scali...

AI Technical Summary

Problems solved by technology

[0006] First, although the current image data is massive, due to the supervision-based learning method, the training data needs to be labeled, so that the current convolutional neural network application has fewer classification objects and is limited to the training library.

[0007] Second, when recognizing more image categories, the scale of the convolutional neural network increases accordingly, and its parameters to be trained increase significantly, which is prone to overfitting

[0008] Third, the convolutional neural network is not very portable to new object recognition tasks, that is, when faced with unknown and complex image recognition objects, it is difficult to satisfy users in terms of training convergence speed and target recognition rate

Images