Unlock instant, AI-driven research and patent intelligence for your innovation.
A multi-scale image target detection method and device
What is Al technical title?
Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
A target detection and multi-scale technology, applied in character and pattern recognition, instruments, computing, etc., can solve difficult multi-scale target detection and recognition problems, achieve the effect of improving computing power, improving efficiency, and high accuracy
Active Publication Date: 2021-02-12
CITY CLOUD TECH HANGZHOU CO LTD
View PDF4 Cites 0 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
[0009] Aiming at the deficiencies of the prior art, the present invention provides a multi-scale image target detection method and device to solve the current target detection system. It is difficult to achieve high accuracy and real-time detection under the condition of high-intensity calculation complexity only by software methods. Technical issues of strong multi-scale target detection and recognition
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more
Image
Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0046] For the processing flow of the above target detection system, its hardware architecture is:
[0047]Multi-CPU (DSP / ARM) architecture, the image preprocessing module function and image processing module function of this architecture are all carried out on the CPU, the image signal conditioning and conversion module and the target signal conditioning and output module functions are realized by a dedicated signal adapter board Interface and protocol conversion, multi-CPU architecture has good versatility and configurability, and is suitable for different processes or tasks of the same task;
Embodiment 2
[0049] For the processing flow of the above target detection system, its hardware architecture is:
[0050] Multi-CPU (DSP / ARM) + multi-ASIC / FPGA / SoC software and hardware collaborative acceleration architecture, using the parallel processing characteristics of the ASIC / FPGA / SoC hardware architecture to realize the low-level processing functions of the image pre-processing module, and also complete data reception and data processing. The CPU completes the control of the ASIC / FPGA / SoC working mode and realizes the advanced processing functions of the image processing module; the frame-by-frame processing and pixel-by-pixel processing of this architecture are realized on the ASIC / FPGA / SoC, and the task is complicated The advanced image processing with high precision and difficult hardware acceleration is realized on the CPU, which achieves a high speed-up ratio while having the whole process of target detection and recognition;
Embodiment 3
[0052] For the processing flow of the above target detection system, its hardware architecture is:
[0053] Multi-CPU+multi-GPU architecture, using the SIMD (Single Instruction Multiple Data) feature of the GPU, the CPU analyzes the operation mode of the image processing module, sends the operation mode instructions to the GPU, and runs the calculation target detection algorithm on the GPU. After the GPU operation is completed, it will process The results are sent back to the CPU. Due to the advantages of GPU in parallel computing and the versatility of CPU control, this architecture has a high acceleration ratio and meets real-time requirements;
[0054] A multi-scale image target detection method, the method specifically includes the following steps:
[0055] Step 1, the image acquisition module collects the original image I, and the collected image is input to the target detection system, and the image preprocessing module of the target detection system performs smoothing p...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More
PUM
Login to View More
Abstract
The invention relates to the technical field of image target detection, and discloses a multi-scale image target detection method and device, including: the detection device divides the original image into multiple scales, namely pixel scale, super pixel scale, and image block scale, and acquires them respectively The initial saliency detection results on each scale; the saliency detection results on multiple scales are used as the feature input of the conditional random field CRF model, and the eigenvalues between adjacent pixels are used as the relationship between random variables to obtain the original image Significant areas of the target area, complete target classification and target positioning; and increase the calculation speed of target detection by accelerating the hardware architecture of the target detection system. The specific acceleration solution is to use multi-CPU architecture to improve the computing power of the system. The efficiency of the system is improved, so that the multi-scale target detection and recognition effect with high accuracy and strong real-time performance can be achieved under the condition of high computational complexity.
Description
technical field [0001] The invention relates to the technical field of image target detection, in particular to a multi-scale image target detection method and device. Background technique [0002] In computer vision, multi-scale is actually the sampling of signals at different granularities. Usually, we can observe different features at different scales to complete different tasks. [0003] Image object detection technology plays an extremely important role and application value in various fields. In the field of personal life, object detection and recognition can be used in home monitoring, smart home, personal electronic devices, etc. For example, at present, smart homes can provide different services based on identifying different family members, and household smart robots can handle various household affairs. In the field of public safety, the security monitoring of airports, subway stations, railway stations, and automatic ticket checking are all very good applicatio...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More
Application Information
Patent Timeline
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to View More 
Login to View More