Object analysis method and device and storage medium

An object analysis and target object technology, applied in the computer field, can solve the problem that the target detection algorithm cannot achieve speed and accuracy, and improve at the same time, and achieve the effect of improving the sensitivity of identifying targets of different scales, improving the accuracy, and speeding up the inference time.
CN113159200AActive Publication Date: 2021-07-23SUZHOU KEDA TECH
4 Cites 2 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SUZHOU KEDA TECH
Publication Date
2021-07-23

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to an object analysis method and device and a storage medium, and belongs to the technical field of computers. The method comprises the steps: acquiring a target image; inputting the target image into a pre-trained image recognition model to obtain an object analysis result of the target image, wherein the object analysis result includes target object position information, target object key point information and target object attribute information; therefore, the problem that an existing target detection algorithm cannot realize simultaneous improvement of speed and accuracy can be solved. Information contained in the obtained feature layer is richer and more comprehensive by using a fusion mode, and the detection accuracy is improved while the sensitivity of identifying targets of different scales is improved; the reasoning time of the fusion part of the feature layer can be shortened by using the depth separable convolution; by integrating the attribute analysis function, one model has three functions of target detection, key point prediction and attribute output at the same time, the video memory occupancy rate of the three functions is reduced, and the running speed of the model in the system is also improved.
Need to check novelty before this filing date? Find Prior Art

Description

【Technical field】

[0001] The present application relates to an object analysis method, device and storage medium, belonging to the technical field of computers. 【Background technique】

[0002] With the development of artificial intelligence technology, more and more target detection algorithms have emerged. Such as anchor-free series algorithms, etc.

[0003] The anchor-free series algorithm avoids the preset of the anchor box, reduces a lot of related calculations, and significantly improves its memory usage and speed. Moreover, there is no presetting of the anchor frame, and the prediction frame returns the target position within the entire feature map range, which is suitable for the detection of target objects of various shapes and sizes.

[0004] However, the existing target detection algorithm will reduce the accuracy of the algorithm while improving the speed of the algorithm, resulting in the problem of inaccurate target detection results. 【Content of invention】 ...

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