摘要：【中文】本发明涉及一种深度二值神经网络训练方法及系统，所述训练方法包括：初始化浮点型的深度神经网络，得到初始化网络模型；基于交替方向乘子法，根据所述初始化网络模型，采用目标传播算法，得到具有二值激活和浮点型权重的优化深度神经网络；基于交替方向乘子法，根据所述优化深度神经网络，得到深度二值神经网络。本发明深度二值神经网络训练方法通过交替方向乘子法优化框架，将权重和激活分别进行二值化，能够减轻同时二值化带来的耦合效应，提高深度二值神经网络的训练效果；采用目标传播算法优化具有二值激活的深度神经网络，能够减小量化过程不可微导致的深度神经网络量化优化困难问题。 【EN】The invention relates to a deep binary neural network training method and a deep binary neural network training system, wherein the training method comprises the following steps: initializing a floating point type deep neural network to obtain an initialized network model; based on an alternating direction multiplier method, obtaining an optimized deep neural network with binary activation and floating point type weight by adopting a target propagation algorithm according to the initialized network model; and obtaining a deep binary neural network according to the optimized deep neural network based on an alternating direction multiplier method. According to the deep binary neural network training method, the frame is optimized through the alternative direction multiplier method, the weight and the activation are respectively binarized, the coupling effect caused by simultaneous binarization can be reduced, and the training effect of the deep binary neural network is improved; the deep neural network with binary activation is optimized by adopting a target propagation algorithm, so that the problem of difficulty in quantization optimization of the deep neural network caused by irreducible quantization process can be solved.

摘要：【中文】本发明涉及图像处理技术领域，具体涉及一种基于一致性负样本的图像目标检测模型训练方法及装置。为了解决现有技术采用固定负样本导致模型难以学习到有判别力的特征的问题，本发明提出一种图像目标检测模型训练方法，该方法包括基于待识别图像中的真实框和预设的初始锚框的重叠度，获取初始图像样本集；根据所述初始图像样本集，通过预设的图像目标检测模型，获取所述与初始锚框对应的预测锚框，基于所述真实框与所述预测锚框的重叠度，获取更新图像样本集；通过所述更新图像样本集训练所述图像目标检测模型。利用本发明的方法和装置能够利用更为全面的信息训练图像目标检测模型。 【EN】The invention relates to the technical field of image processing, in particular to a method and a device for training an image target detection model based on a consistency negative sample. In order to solve the problem that discriminative features are difficult to learn by a model due to the adoption of a fixed negative sample in the prior art, the invention provides an image target detection model training method, which comprises the steps of obtaining an initial image sample set based on the overlapping degree of a real frame and a preset initial anchor frame in an image to be recognized; acquiring the prediction anchor frame corresponding to the initial anchor frame through a preset image target detection model according to the initial image sample set, and acquiring an updated image sample set based on the overlapping degree of the real frame and the prediction anchor frame; and training the image target detection model through the updated image sample set. By using the method and the device, the image target detection model can be trained by using more comprehensive information.

摘要：【中文】本发明涉及图像处理技术领域，具体涉及一种基于无标签数据的神经网络模型量化方法及装置。为了解决现有技术需要依赖训练集才能实现对神经网络压缩的问题，本发明提出一种基于无标签数据的神经网络模型量化方法，包括基于第一预设量化位宽对原始神经网络模型的权重进行量化，获取第一量化神经网络模型；基于原始神经网络模型的输出和第一量化神经网络模型的输出，对第一量化神经网络模型的每一层不同通道的量化位宽进行优化，获取第二量化神经网络模型；基于原始神经网络模型的特征和第二量化神经网络模型的特征，对第二量化神经网络模型的权重进行优化，获取目标神经网络模型。本发明的方法能够通过少量无标签数据对神经网络模型进行量化。 【EN】The invention relates to the technical field of image processing, in particular to a neural network model quantification method and device based on label-free data. In order to solve the problem that the neural network compression can be realized only by depending on a training set in the prior art, the invention provides a neural network model quantization method based on label-free data, which comprises the steps of quantizing the weight of an original neural network model based on a first preset quantization bit width to obtain a first quantized neural network model; optimizing the quantization bit width of each layer of different channels of the first quantization neural network model based on the output of the original neural network model and the output of the first quantization neural network model to obtain a second quantization neural network model; and optimizing the weight of the second quantitative neural network model based on the characteristics of the original neural network model and the characteristics of the second quantitative neural network model to obtain the target neural network model. The method of the invention can quantify the neural network model through a small amount of label-free data.

摘要：【中文】本发明属于计算机视觉领域，具体涉及了一种基于分块分类深度学习的篡改图像检测方法及系统，旨在解决现有技术篡改图像检测准确率和定位精度尚达不到实用要求的问题。本发明方法包括：分别对待检测图像灰度化后分块分类以及设定算子的边缘特征计算；基于分块分类结果，进行待检测图像及特征图的分块分类；通过图像块篡改检测模型获取分类图像块的检测结果、特征图像块的检测结果并进行加权；结合的图像块的标记，进行待检测图像的篡改区域标记。本发明对图像分块分类，分别训练篡改检测模型，检测结果更准确、篡改区域定位更精确，并将未分块的图像边缘直接定义为其最邻近图像块的篡改检测结果，提高检测效率。 【EN】The invention belongs to the field of computer vision, and particularly relates to a method and a system for detecting a tampered image based on block classification deep learning, aiming at solving the problem that the detection accuracy and the positioning accuracy of the tampered image in the prior art can not meet practical requirements. The method comprises the following steps: respectively classifying blocks of the image to be detected after graying and calculating the edge characteristics of a set operator; based on the block classification result, performing block classification on the image to be detected and the feature map; obtaining the detection result of the classified image block and the detection result of the characteristic image block through the image block tampering detection model and weighting; and marking the tampered area of the image to be detected by combining the marks of the image blocks. The method classifies the image blocks, respectively trains the tampering detection models, has more accurate detection results and more accurate positioning of tampered areas, directly defines the image edge which is not subjected to blocking as the tampering detection result of the nearest image block, and improves the detection efficiency.