A method for detecting advertising image material
A detection method and material technology, applied in the field of computer vision, can solve the problem of low detection accuracy of advertising image materials, and achieve the effect of improving accuracy, improving ability, and fine detection.
Active Publication Date: 2021-11-30
UNIV OF SCI & TECH BEIJING
View PDF8 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] The technical problem to be solved by the present invention is to provide a detection method for advertising image material to solve the problem of low detection accuracy of advertising image material caused by the general target detection method in the prior art
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will describe in detail with reference to the drawings and specific embodiments.
[0042] Aiming at the problem of low detection accuracy of advertisement image material caused by the existing general object detection method, the invention provides an advertisement image material detection method.
[0043] Such as figure 1 As shown, the advertising image material detection method provided by the embodiment of the present invention includes:
[0044]S101. Obtain an advertising image material detection training set;
[0045] S102. Construct a multi-scale instance segmentation network, and use the acquired images in the training set to train the multi-scale instance segmentation network, wherein the multi-scale instance segmentation network is used to extract pixel-level semantic information and Connect information between pixels, and 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 provides a method for detecting an advertisement image material, which can realize fine detection of the advertisement image material. The method includes: obtaining a training set for detecting advertisement image materials; constructing a multi-scale instance segmentation network, and using the acquired images in the training set to train the multi-scale instance segmentation network, wherein the multi-scale instance segmentation network is used to extract The semantic information and connection information are processed by spatial pyramid pooling with holes to obtain material features of different scales. Semantic prediction is performed based on the obtained material features of each scale, and the obtained material features are cascaded to obtain a connection representing the connection relationship between pixels. Predict and represent the contour prediction of the boundary of the advertising image material, form a loss function of semantic loss, connection loss and contour loss, jointly optimize the multi-scale instance segmentation network, connect the connection prediction result with any semantic prediction result, and output the image ad image assets. The present invention relates to the field of computer vision.
Description
technical field [0001] The invention relates to the field of computer vision, in particular to a method for detecting advertising image materials. Background technique [0002] Advertising images play an increasingly important role in people's daily life. Advertisement image material refers to those image areas in the advertisement image that contain effective product information, which is distinguished from background content such as text and tables in the advertisement image. Although advertising image materials have broad application value in the field of advertising, such as image materials for dynamic video advertisements, data for training depth models, etc., there is currently a lack of effective methods for extracting advertising image materials. Although the general object detection method (Liu, Li, et al. "Deep learning for generic object detection: A survey." arXiv preprint arXiv:1809.02165 (2018)) can also be used for creative extraction, if directly applied to ...
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
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06N3/04G06T7/12G06T7/13G06T7/181
CPCG06T7/13G06T7/12G06T7/181G06N3/045G06F18/214
Inventor 殷绪成徐奔杨春
Owner UNIV OF SCI & TECH BEIJING