A system and method for detecting violent events in images based on anchor nodes
An event detection and anchor node technology, applied in computer parts, instruments, computing, etc., can solve problems such as being easily affected by noise, unable to achieve real-time effects, etc., and achieve the effect of improving efficiency and accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0068] The scenario described in this embodiment is a distributed application scenario that does not require real-time processing. For example, after cameras on several streets capture images, the images are transmitted to a host to store image files, and the host processes the corresponding images of a small number of cameras. image and obtain the prediction results, and at the same time temporarily store the results locally, upload the temporarily stored prediction results to their respective databases every few hours, and provide the client with the query function through the reverse proxy server, The client does not feel the existence of multiple databases, and can directly access the data of a large number of cameras, such as figure 2 The specific steps are as follows:
[0069] Before the system is deployed, collect a large number of image data sets for training on the Internet or other means, organize the data sets, accurately classify all image data and add correspondi...
Embodiment 2
[0074] The scene described in this embodiment is: some special places that need to centrally process images in real time. Since the area is generally small and there are few cameras as a whole, there is a high demand for real-time monitoring of violent events. Therefore, centralized real-time prediction of these images is required. Each video streaming server is only connected to a small number of cameras. These servers are only responsible for image storage and perform a small amount of processing to obtain hash codes. These hash codes will be marked with the source and sent to the same detection server. Stored in the database, the client will poll the database and issue a warning when a violent event is queried. refer to image 3 ,Specific steps are as follows:
[0075] Before the system is deployed, collect a large number of image data sets for training on the Internet or other means, organize the data sets, accurately classify all image data and add corresponding labels t...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More