首页>
外国专利>
NEURAL NETWORK-BASED MULTILAYER IMAGE FEATURE EXTRACTION MODELING METHOD AND DEVICE AND IMAGE RECOGNITION METHOD AND DEVICE
NEURAL NETWORK-BASED MULTILAYER IMAGE FEATURE EXTRACTION MODELING METHOD AND DEVICE AND IMAGE RECOGNITION METHOD AND DEVICE
展开▼
机译：基于神经网络的多层图像特征提取建模方法及装置和图像识别方法及装置
展开▼
页面导航
摘要
著录项
相似文献
摘要
A neural network-based multilayer image feature extraction modeling method and device, the method comprising: obtaining a first image, a second image, a first category of the first image, and a second category of the second image from a training set of a preset application scenario (S140); determining a global loss/cost function value according to the first image, the first category, the second image, and the second category (S160); training a multilayer image/object/authentication neural network on the training set according to the global loss/cost function value and a training parameter (S170); and testing the multilayer image/object/authentication neural network by means of a testing set of the preset application scenario, determining testing precision according to the testing result, and determining a target multilayer image/object/authentication feature extraction model according to the testing precision and the multilayer image/object/authentication neural network (S180). The method and device can achieve the beneficial effect of improving image recognition precision when performing image recognition in an image recognition application scenario using an image feature model obtained by modeling. Also provided are an image recognition method and device.
展开▼
