The invention discloses a generative-adversarial-network-based blurred face reconstruction method and system. The face reconstruction method comprises: acquiring a training sample and to-be-re-reconstructed data; carrying out generative adversarial network training on two convolutional neural networks by the training sample to obtain an optimized-target-based generative network model; and acquiring a blurred face from the to-be-re-reconstructed data, inputting the blurred face into the generative network model to carry out de-blurring reconstruction of the face, thereby obtaining a clear face.The face reconstruction method and the system have the following beneficial effects: a phenomenon that the image obtained by using an interpolation method is too smooth is avoided; the image recovered by the model becomes clear and more high-frequency details are provided; because no frequency domain method is needed, a phenomenon that the frequency-domain data are lack of correlation is avoided;and the blurred face or shielded face can be reconstructed rapidly and the definition and similarity are high. In addition, the face reconstruction system is composed of an acquisition unit, a modelgeneration unit, and a face reconstruction unit; and the face reconstruction system has the same beneficial effects as the face reconstruction method.

Generative-adversarial-network-based blurred face reconstruction method and system

The invention discloses a workpiece surface defect detecting method based on convolutional neural networks. The method comprises the following steps: each surface image of a workpiece is acquired; a convolutional neural network is constructed; a training sample is constructed; a parameter of the convolutional neural network is trained; an image of a workpiece to be measured is inputted into the trained convolutional neural network, and a defect area marking image of a pixel grade is obtained. The input of the convolutional neural network for defect detection is a pretreated workpiece originaldrawing, after neural network calculation, a defect labeled graph with the size which is identical to the size of an original graph is outputted, and by means of analysis of the defect labeled graph,better obtaining position, size, shape and other important information of the defect areas are obtained. Compared with the traditional defect detection methods, the method can be used for processing image data of large scale, reducing massive workload with manual design of characteristics, and at the same time processing a plurality of different defect classes and obtaining defect area informationof pixel grade.

Workpiece surface defect detecting method based on convolutional neural networks

The invention discloses a WGAN model method based on a depth convolution nerve network, and belongs to the deep learning nerve network field; the method comprises the following steps: S1, forming a Wasserstein generative confrontation network WGAN model; S2, forming a discriminator with the depth convolution nerve network structure; S3, forming a generator with a transposition convolution nerve network structure; S4, employing a Wasserstein distance loss function as a discriminator loss function; S5, preparing a data set so as to train the formed network. The method creatively proposes the depth convolution nerve network and WGAN combined forming mode according to the generative confrontation network model features, forms the generator and the discriminator as the depth convolution nerve network mode, and uses the WGAN loss function mode; the method can learn image features in the training process, and can use the loss functions to reflect the generative image quality.

WGAN model method based on depth convolution nerve network

The invention relates to an image single classification method based on a generative confrontation network. The image single classification method based on a generative confrontation network includesthe steps: constructing a generator in the generative confrontation network by means of a dense connection block structure; constructing a discriminator in the generative confrontation network; inputting positive sample training data, and using a gradient punishment algorithm to train the generative confrontation network; according to the classification effect of the model on the verification setduring the training process, adjusting the network parameters, and using the Early Stopping strategy to find the classification optimal iteration number of the model; and after the model training is completed, using the discriminator in the generative confrontation network to test the test set data, and determining the classification effect of the model through a classification recall index CRI determination model. The image single classification method based on a generative confrontation network can automatically generate a negative sample set, and can solve the problem that in a current single classification method, artificial construction of a negative sample data set is likely to cause over-fitting of the classifier.

Image single classification method based on generative confrontation network

The invention discloses a data sample tag processing method and apparatus. The method comprises: an indeterminate tag sample set and a determined tag sample set are obtained; iterative processing is carried out by the following steps until the learning tag accuracy reaches a preset requirement: the current indeterminate tag sample set and determined tag sample set are combined and training is carried out to obtain a prediction model; indeterminate tag samples are inputted into the prediction model an prediction value outputted by the model is determined as a learning tag of indeterminate tag samples; a sampling sub set with a learning tag is obtained and the learning tag of the sampling sub set is checked to obtain an accuracy rate of the learning rate; if the accuracy rate of the learningtag does not reach the preset requirement, the tag checking result is determined as a marking result of the sampling sub set and the sampling set is transferred from the indeterminate tag sample setto the determined tag sample set; and after ending of iteration, the current learning tag of the indeterminate tag sample set is determined as a marking result.

Data sample tag processing method and apparatus

The invention relates to a pest and disease image generation method based on a generative adversarial network In the prior art, sampling images of pest and disease images are less By using the method of the invention, the above defect is overcome The method comprises the following steps of collecting and preprocessing trained images; based on a deep-convolution neural network model, constructing a discrimination network and a generation network; training the discrimination network and the generation network; and according to the trained generation network, generating the pest and disease images In the invention, according to a few of existing pest and disease images, a lot of pest and disease images which are similar to a reality are generated, a sample image is provided for pest and disease image identification, and problems that the pest and disease images in an actual field are less and acquisition cost is high are solved

Pest and disease image generation method based on generative adversarial network

The invention relates to a pest image classification method based on a grading-prediction convolutional neural network. Compared to the prior art, by using the method of the invention, a defect that a pest image classification correct rate is low is overcome. The method comprises the following steps of collecting and preprocessing training images; marking image sample data; training a classification model based on the grading-prediction convolutional neural network; preprocessing images to be tested; and based on a grading model, automatically carrying out pest image classification. In the invention, a grading prediction framework is adopted; a segmentation result of the images is predicted firstly and then an integral image is combined; and final classification prediction is performed.

Pest image classification method based on grading-prediction convolutional neural network

The invention relates to a deep learning technology based soil near infrared spectroscopic analysis prediction method. Compared with the prior art, the defect that large-scale comprehensive analysis cannot be conducted on soil constituents is solved. The deep learning technology based soil near infrared spectroscopic analysis prediction method comprises the following steps of obtaining and preprocessing of a training sample, establishment of a deep learning based prediction model, obtaining and preprocessing of a testing sample, inputting of spectroscopic data of the preprocessed testing sample into the constructed prediction model to complete analytical prediction of the soil constituent content of the testing sample, and by taking Mel domain smoothing band coefficients as parameters, changingspectroscopic data oftesting soil sample intotwo-dimensional characteristics to be inputted into a trained convolutional neural network for analytical prediction of the soil constituent content. The deep learning technology based soil near infrared spectroscopic analysis prediction method performs soil near infrared spectroscopic analysis prediction based on a structural model of the convolutional neural network and improves the accuracy of the near infrared spectroscopic prediction of main soil constituents and the robustness of the model.

Deep learning technology based soil near infrared spectroscopic analysis prediction method

The invention relates to a pest image identification method based on a multi-space convolution neural network, and solves the shortcomings of low image identification rate and poor robustness compared with the prior art. The method comprises the following steps of collecting and pre-processing training images, constructing a multi-scale MS-CNN network model and a multi-core classification model, collecting and pre-processing image to be tested, taking a test sample as input and training the multi-core model and the MS-CNN model, and conducting automatic identification of pest images after the training. According to the pest image identification method based on the multi-space convolution neural network, the accuracy rate of pest identification is improved, the robustness of a pest identification algorithm is enhanced and the practical application level is achieved.

Pest image identification method based on multi-space convolution neural network

The invention relates to a disease image identification method suitable for multi-size picture information, and overcomes the defects of low identification rate and poor robustness due to that the resolutions and sizes of disease images are different as compared with the prior art. The method of the invention includes the steps of collecting and pre-processing training images; constructing and training an MCNN network model; collecting and pre-processing images under test, shooting disease images to be identified using an acquisition device, and performing histogram equalization of the disease images under test to obtain a test sample; and inputting the test sample into the trained MCNN network model for automatic identification of the disease images. The method of the invention can directly extract features of the images with different resolutions and sizes, without unifying the size of the images before the identification, and perform classification prediction.

Song Liangtu

Papers

Pest and disease image generation method based on generative adversarial network

Pest image classification method based on grading-prediction convolutional neural network

Deep learning technology based soil near infrared spectroscopic analysis prediction method

Pest image identification method based on multi-space convolution neural network

Disease image identification method suitable for multi-dimensional picture information