Unlock instant, AI-driven research and patent intelligence for your innovation.

Behavior recognition method and device based on voting mechanism and storage medium

A voting mechanism and recognition method technology, applied in biometric recognition, character and pattern recognition, computer parts and other directions, can solve the problems of different video lengths, poor behavior recognition accuracy, and poor practicability, and achieve recognition results. accurate effect

Pending Publication Date: 2022-06-17
PING AN TECH (SHENZHEN) CO LTD
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Existing behavior recognition faces great challenges in specific applications. The main difficulty lies in how to accurately and strictly cut out the segment from the beginning to the end of the behavior from a long video; in addition, due to the different lengths of videos , and there are many factors such as multi-scale, multi-target, and camera movement in the video in an open environment, resulting in poor accuracy of behavior recognition and poor practicability.
[0003] For example, in the existing smart city

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
  • Behavior recognition method and device based on voting mechanism and storage medium
  • Behavior recognition method and device based on voting mechanism and storage medium
  • Behavior recognition method and device based on voting mechanism and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0051]应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。

[0052]本发明提供一种基于投票机制的行为识别方法。参照图1所示，为本发明一实施例提供的基于投票机制的行为识别方法的流程示意图。该方法可以由一个装置执行，该装置可以由软件和 / 或硬件实现。

[0053]在本实施例中，基于投票机制的行为识别方法包括：

[0054]S100：获取带有样本行为的目标视频的第一时域段。

[0055]其中，目标视频可通过指定场所安装的监控设备进行获取，或者从存储数据库中直接进行视频调取，视频中的样本行主要指目标行为，其为可包括吸烟行为、扔垃圾行为、驾驶行为等多种类型的目标行为，样本行为主要是指正常发生该行为时的行为，并不包括特殊情况下的其他无关行为，例如，针对扔垃圾的样本行为，主要是从用户进入垃圾桶一定区域内发生的扔垃圾行为，而环卫工人的正常作业，以及在垃圾桶附近发生的谈话等，均不属于目标行为。此外，对应的第一时域段为该样本行为自起始至结束的一个粗略的时域段，针对该时域段内的视频进行分析，行为的识别准确度更高，此外能够避免对所有视频段进行识别，可以减少计算压力。

[0056]其中，该步骤中获取第一时域段可进一步包括：

[0057]S110：基于检测模型获取所述样本行为的行人框以及目标物体的物体框；

[0058]S120：获取所述行人框的第一位置坐标信息，以及所述物体框的第二位置坐标信息；

[0059]S130：基于所述第一位置坐标信息和所述第二位置坐标信息，确定所述第一时域段。

[0060]其中，确定行人框的第一位置坐标信息以及物体框的第二位置坐标信息之后，可以根据坐标信息判断二者之间的距离是否足以触发相应的样本行为，如果二者之间的坐标信息距离较远，则可确认没有发生样本行为的可能，直至二者的坐标信息达到一定的要求，此时视频对应的时间点可以认定为第一时域段的起始时间，同理，当二者之间的距离又超过距离要求时，可据此确定第一时域段的终止时间。

[0061]作为具体示例，可采用GIoU(Geberalized Intersection over Union，联合体上几何化交叉)来确定样本行为的初始值结束的粗略范围，采用GloU可以避免视频摄像头远近和高低的影响，同时能够较好的反映出行人和目标物体之间的位置关系。

[0062]具体地，如下图2所述，左下角框A表示行人框，右上角框B表示...

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

No PUM Login to View More

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a voting mechanism-based behavior recognition method, which comprises the following steps of: acquiring a first time domain segment of a target video with a sample behavior; based on the first time domain segment, acquiring an amplified video segment corresponding to the target video; respectively inputting the amplified video segments into a preset neural network model for training until the neural network model is trained into a behavior recognition model; behavior recognition is carried out on behaviors in a to-be-recognized video based on the behavior recognition model, and a corresponding recognition result is obtained; and voting the recognition result, and determining a final behavior result of the to-be-recognized video based on the voting result. The accuracy based on behavior recognition can be improved.

Description

technical field [0001] The present invention relates to artificial intelligence technology, in particular to a method, device, electronic equipment and computer-readable storage medium for behavior recognition based on a voting mechanism. Background technique [0002] At present, behavior recognition for videos mainly refers to judging the types of behaviors in the pre-segmented sequences in the time domain, that is, "reading behaviors". Existing behavior recognition faces great challenges in specific applications. The main difficulty lies in how to accurately and strictly cut out the segment from the beginning to the end of the behavior from a long video; in addition, due to the different lengths of videos , and there are many factors such as multi-scale, multi-target, and camera movement in the video in an open environment, resulting in poor accuracy of behavior recognition and poor practicability. [0003] For example, in the existing smart city construction, garbage cla...

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
no application Login to View More
IPC IPC(8): G06V40/20G06V40/10G06V20/40G06V10/774G06V10/80G06V10/82G06K9/62G06N3/04
CPCG06N3/045G06F18/25G06F18/259G06F18/214
Inventor 陈丹陆进刘玉宇肖京
Owner PING AN TECH (SHENZHEN) CO LTD