The invention provides a neural network model processing method, a device and a terminal, which relate to the technical field of the neural network. The method comprises the following steps: for eachconvolution layer of the neural network model, the parameter encoded characters of the current convolution layer are read from a memory, and the parameter encoded characters are decoded to obtain theparameters of the current convolution layer. According to the parameters of the current convolution layer, the convolution calculation of the current convolution layer is carried out, and the calculation result is obtained. The result of the calculation is encoded and stored in a memory as one of the parameter encoded characters of the next convolution layer of the current convolution layer. A method for processing a neural network model provided by an embodiment of that present invention Devices and terminals. The storage capacity of the neural network model is effectively reduced, which provides the possibility for the terminal to deploy the large-scale neural network model, at the same time, the memory reading times are saved, the interaction time between the memory and the neural network model is reduced, the reasoning process of the neural network model is further accelerated, and the computing speed of the neural network model is also improved.

Neural network model processing method, device and terminal

An apparatus for a convolutional neural network includes a processing engine two-dimensional array. The processing engine two-dimensional array is configured to receive the input data and the weight data and comprises an array of N processing engine rows and M processing engine columns, and N and M are both positive integers larger than or equal to 2. And the processing engine two-dimensional array performs convolution operation on the input data and the weight data and then outputs an intermediate result. And the processing engine of the processing engine two-dimensional array is configured to carry out pooling processing after the convolution operation. The processing engine two-dimensional array may also have a full-connection structure, and an intermediate result is superimposed withinthe processing engine two-dimensional array. The apparatus can process the input tensor data at a high speed, and can flexibly cope with the input tensor data of different dimensions.

An apparatus for a convolutional neural network

A hardware accelerator for neural network convolution operation comprises a processing engine matrix, a weight data buffer, a threshold value data buffer, an input data buffer and a control module. The control module reads input data from the input data buffer and transmits the input data to a processing engine matched with the input data in the processing engine matrix; the control module reads the weight data from the weight data buffer and configures the weight data to a processing engine matched with the weight data; and the control module reads threshold value data from the threshold value data buffer and configures the threshold value data to each processing engine. Each processing engine carries out convolution operation on the matched input data and weight data to obtain an intermediate result, the intermediate result is compared with threshold value data, and the intermediate result or a regularization result obtained after regularization is selected to be output according toa comparison result. The hardware accelerator realizes high-speed processing of input tensor data and can flexibly cope with input tensor data of different dimensions.

Hardware accelerator for neural network convolution operation

The invention discloses a fixed-point acceleration method for data. The method comprises the following steps: acquiring test data and first model parameters; inputting the test data and the first model parameters into a fixed-point algorithm model needing fixed-point parameter adjustment to obtain a first characteristic value; Converting the data format of the first characteristic value into a floating point format from a fixed point format, and obtaining a second characteristic value represented by the floating point format; and performing fixed-point operation on the second characteristic value by using a GPU acceleration mechanism to obtain output data. The invention further provides a fixed-point acceleration device, electronic equipment and a computer storage medium. According to theinvention, the calculation of the fixed-point algorithm model can be accelerated, and the fixed-point iteration period is accelerated.

Data fixed-point acceleration method and device, electronic equipment and storage medium

A neural network inference structure optimization method, comprising: when an Mth network layer and a (M+2)th network layer of a neural network inference structure are both normalization layers, the (M+1)th network layer is a convolutional layer or a fully connected layer, and an output of the (M+1)th network layer is only connected to the (M+2)th network layer (501), invoking a first preset algorithm to process the (M+1)th network layer, so as to merge the (M+2)th network layer into the (M+1)th network layer to obtain a first optimized network layer of the (M+1)th network layer (502); and invoking a second preset algorithm to process the first optimized network layer of the (M+1)th network layer, so as to merge the Mth network layer into the first optimized network layer of the (M+1)th network layer (503). The present invention can reduce the calculation amount and processing delay in neural network inference to the greatest extent, achieving the purpose of improving the inference efficiency of a neural network model.

Neural network inference structure optimization method and device

The embodiment of the invention discloses an optimization method and device of a convolutional neural network (CNN) and a computer storage medium. The method includes: constructing the convolutional neural network, wherein the convolutional neural network includes at least four network layers of an image input layer, at least one convolutional layer, at least one pooling layer and at least one fully connected layer; decreasing the number of convolution kernels in the CNN when the number of to-be-detected objects is less than a preset threshold value; dividing an input image of the image inputlayer into at least one memory data segment, which is stored by utilizing continuous memory, according to a set border determination strategy, and utilizing a set continuous memory copy function to carry out data copying on each memory data segment; and merging original parameters in a batch normalization (BN) layer and parameters of the convolutional layer or the fully connected layer according to the set merging strategy, and using merged parameters as new parameters of the batch normalization layer, wherein the batch normalization layer is after the convolutional layer or the fully connected layer. Calculation consumption in carrying out detection through the CNN is reduced.

Optimization method and device of convolutional neural network (CNN) and computer storage medium

The embodiment of the invention discloses an image processing method, equipment and a storage medium. The method can comprise the steps that at least one hand object in detection image frames of a video is detected based on set hand training data and a convolutional neural network (CNN) model, and description property values of all the hand objects are acquired; and when the description property values of all the hand objects meet set triggering conditions, hand objects in a set quantity of tracking image frames following the detection image frames are tracked based on set object features anda tracking algorithm. According to the technical scheme in the embodiment, by fusing a detection algorithm and a tracking algorithm, the number of times of convolution calculation needed during hand detection is reduced, therefore, calculation cost generated during hand detection is reduced, and time consumed in the detection process is shortened.

Image processing method, equipment and storage medium

Handle everyday research tasks with reliable, citation-backed results

Your personal Research Agent to handle research tasks with citation-backed results

Popular Tasks used by Researchers

How can I help with your research?

Meet SciSpace

Get more enhanced response by uploading the PDFs you want me to reference.

No relevant PDFs in your library

SciSpace is the AI research assistant for academics. Run systematic literature reviews on 280M+ papers, and write papers with cited sources. Trusted by 1M+ students, PhDs & researchers.

SciSpace | AI for Research

Analyze PDFs

Code & Manuscripts

Funding & Grants

Literature & Patents

Medical & Clinical Data

Systematic Review

Visualize & Present

Web & Data

Build a Google Scholar-like website for your research.

Build a website

Create charts and images for your research

Create a Chart

Write a paper for submission to a journal

Draft a manuscript

Patent Search

Design eye-catching scientific posters in minutes.

Scientific Poster Generation

Systematic Literature Review

One task is running at the moment. Your messages will be shown right after.

Drag and drop or click here to browse

Loved by <highlight>1 million+</highlight> researchers

Extract a list of specific topics and their sources from unstructured text

Topics

Compare and analyze relevant papers that matches with your search

Papers

Get insights from PDFs and bookmarked papers from your library

My library

Recent searches

Try searching for:

Catch AI-generated content in scholarly and non-scholarly content

{ai} Detector

Ai Writer

Get PDF Summaries, highlighted text explanations 

Chat with PDF

Effortlessly create in-text citations and bibliographies in APA and 2,500 other formats

Citation generator

Get explanations, summaries, and answers on academic papers

Ease up your research workflow with {scispace}'s cohort of exciting AI tools

Elevate your academic writing skills and convey your ideas the way you want

Paraphraser

Explore our range of reading and writing tools

Your file is being prepared and should be ready in a few minutes. If it's a large file, it might take a bit longer. You can close this window, and we'll email you the file when it's done.

You have reached a maximum limit of <strong>{limit}</strong> columns in the table. Remove at least <strong>1</strong> column to add or create another one.

Wang Xinbo

Author Tools

Chat about Author

Papers

Optimization method and device of convolutional neural network (CNN) and computer storage medium

Image processing method, equipment and storage medium