Video behavior classification method, model training method, device, medium and equipment
A classification method and behavior technology, applied in the field of image processing, can solve problems such as low computing efficiency and the inability of video behavior classification algorithms to be deployed on edge devices, etc., to achieve the effects of enhancing robustness, reducing computing complexity, and improving computing efficiency
Active Publication Date: 2022-05-10
小视科技(江苏)股份有限公司
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] This application provides a video behavior classification method, model training method, device, medium and equipment, which are used to solve the problem that the current mainstream video behavior classification algorithm cannot be deployed on the edge device and the calculation efficiency is low
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
[0072] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the following will further describe the embodiments of the present application in detail in conjunction with the accompanying drawings.
[0073] Video behavior classification refers to classifying the behavior of the target object in the video stream. The specific categories can be normal state, fighting, sitting down, standing up, running, playing football, stealing, robbing, etc., which are not limited in this embodiment.
[0074] In this embodiment, the computer device needs to train the model first, and then use the model to classify video behaviors. The training process of the model will be described below.
[0075] Please refer to figure 1 , which shows a method flowchart of a model training method provided by an embodiment of the present application, and the video behavior classification method can be applied to a computer device. The model tra...
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 application discloses a video behavior classification method, model training method, device, medium and equipment, belonging to the technical field of image processing. The method includes: extracting the i-th video frame sequence from the i-th video segment in the video stream; sampling the i-th video frame sequence according to a first sampling rule to obtain the i-th image sequence; Arrange positions in the i image sequence, add a digital watermark to each image in the i image sequence respectively; according to the image mosaic template and digital watermark, sequentially splicing each image in the i image sequence to obtain a mosaic Image: Input the stitched image into the trained model, and generate the classification result of the i-th video segment according to the category predicted by the model. The model is created based on a two-dimensional convolutional neural network. The model created based on the two-dimensional convolutional neural network in this application can be deployed in edge devices, and the robustness and computational efficiency of the model are high.
Description
technical field [0001] The present application relates to the technical field of image processing, in particular to a video behavior classification method, model training method, device, medium and equipment. Background technique [0002] Video behavior classification has a wide range of applications in video understanding, behavior recognition and other fields. The current mainstream method is to use 3D convolutional neural network (Convolutional Neural Networks, CNN) or Vision Transformer (VisionTransformer) and other networks to classify video behavior. Although these networks have certain advantages in classification accuracy, there are also parameters and calculations. Due to problems such as large volume, immature and common algorithms, it is prone to the problem that the operator does not support when deploying to the edge device, so that the deployment cannot be realized. In addition, the large amount of parameters and calculation will also affect the calculation ef...
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): G06V20/40G06V10/774G06V10/764G06V10/82G06K9/62G06T1/00G06T3/40G06N3/04G06N3/08
CPCG06T1/0021G06T3/4038G06N3/08G06N3/045G06F18/214G06F18/24
Inventor 杨帆冯帅白立群胡建国
Owner 小视科技(江苏)股份有限公司