技术领域technical field
本发明涉及计算机视觉领域及视频图像处理领域，尤其涉及一种基于Kinect摄像机的驾驶行为识别方法。The invention relates to the field of computer vision and video image processing, in particular to a driving behavior recognition method based on a Kinect camera.
背景技术Background technique
当前，港口驾驶员的违规行为屡次直接导致港口重大事故，例如，驾驶员的一个违规行为往往可能导致整个生产线停产，严重时会造成设备损毁和人员伤亡。通过调研发现，驾驶员的违规行为主要集中在，操作员注意力不集中，驾驶过程中出现玩手机、接听电话或双手长时间脱离操作杆的违规驾驶行为。有鉴于此，对驾驶员的当前驾驶行为进行实时识别和预警，对提升港口作业的安全可靠性具有现实价值和重要作用。At present, the violations of port drivers often directly lead to major port accidents. For example, one violation of the driver may often lead to the shutdown of the entire production line, and in severe cases, equipment damage and casualties may be caused. Through investigation, it is found that the driver's violations are mainly concentrated, the operator is inattentive, and the illegal driving behavior of playing with mobile phones, answering the phone or taking his hands off the operating lever for a long time during driving occurs. In view of this, real-time identification and early warning of the driver's current driving behavior has practical value and important role in improving the safety and reliability of port operations.
在现有技术中，港口虽存在监控系统来监督驾驶员依照规定姿势进行安全驾驶，但同样需要人工监控。然而这也会带来另外一个问题，即，监控人员的注意力不集中，出现违规驾驶行为时的监控视频不能被监控人员及时捕捉时，仍然不能有效地制止事故的发生，更无法对危险驾驶行为进行预判，因此需要一种无须人为干预的驾驶员操作行为识别方法。In the prior art, although there is a monitoring system in the port to supervise the driver to drive safely according to the prescribed posture, manual monitoring is also required. However, this also brings another problem, that is, the monitoring personnel are not attentive, and when the monitoring video of illegal driving behavior cannot be captured by the monitoring personnel in time, the accident still cannot be effectively prevented, let alone dangerous driving. Therefore, there is a need for a driver operation behavior recognition method without human intervention.
另一方面，行为识别涉及计算机识别和模式识别等领域，是一个重要且又有挑战性的课题。行为识别即基于视频或图像，提取有效表达行为的特征信息，并进一步对特征进行识别从而得到人体的具体行为。传统的识别方法是对图片进行特征提取从而进行识别，例如梯度直方图(HOG)、光流直方图(HOF)、尺度不变特征变化(SIFT)，但往往会受到复杂环境的影响，如光照干扰等。由上述可知，现有的行为识别主要存在两方面的问题：一是如何选取强有力的特征对人体行为进行描述，二是如何对动态时间进行建模，既不损失有效信息同时还能最大限度地利用信息进行识别。On the other hand, behavior recognition involves fields such as computer recognition and pattern recognition, and is an important and challenging subject. Behavior recognition is to extract the feature information that effectively expresses the behavior based on the video or image, and further recognize the feature to obtain the specific behavior of the human body. The traditional recognition method is to perform feature extraction on the image for recognition, such as histogram of gradient (HOG), histogram of optical flow (HOF), scale-invariant feature change (SIFT), but it is often affected by complex environments, such as lighting. interference, etc. It can be seen from the above that there are two main problems in the existing behavior recognition: one is how to select powerful features to describe human behavior, and the other is how to model the dynamic time without losing effective information and at the same time maximizing human behavior. use information for identification.
有鉴于此，如何设计一种针对驾驶员的驾驶行为识别方法，从而克服现有技术中的上述缺陷或不足，是业内相关技术人员需要解决的一项课题。In view of this, how to design a driving behavior recognition method for drivers, so as to overcome the above-mentioned defects or deficiencies in the prior art, is a subject that needs to be solved by the relevant technical personnel in the industry.
发明内容SUMMARY OF THE INVENTION
针对现有技术的驾驶行为识别和监控时所存在的上述缺陷，本发明提供了一种基于Kinect摄像机的驾驶行为识别方法。Aiming at the above-mentioned defects in the prior art driving behavior identification and monitoring, the present invention provides a driving behavior identification method based on a Kinect camera.
依据本发明的一个方面，提供了一种基于Kinect摄像机的驾驶行为识别方法，包括以下步骤：According to an aspect of the present invention, there is provided a driving behavior recognition method based on a Kinect camera, comprising the following steps:
a)利用Kinect摄像机采集的视频帧对前景目标驾驶员进行骨架追踪，并定位驾驶员的骨架数据；a) Use the video frames collected by the Kinect camera to track the skeleton of the foreground target driver, and locate the skeleton data of the driver;
b)根据所述驾驶员的骨架数据，获取所述视频帧的每一帧中的驾驶员的多个关节点坐标值并对其进行归一化处理，其中所述关节点坐标值之间的内部关系用于表征所述驾驶员的驾驶行为；b) According to the skeleton data of the driver, obtain and normalize a plurality of joint point coordinate values of the driver in each frame of the video frame, wherein the difference between the joint point coordinate values Internal relationships are used to characterize the driver's driving behavior;
c)对所述关节点坐标值进行特征提取，生成多个单帧骨骼特征向量；以及c) performing feature extraction on the joint point coordinate values to generate a plurality of single-frame skeleton feature vectors; and
d)将所述多个单帧骨骼特征向量送入预设的深度学习模型进行识别，根据识别结果判断当前驾驶状态是标准驾驶行为或违规驾驶行为。d) Sending the plurality of single-frame skeleton feature vectors into a preset deep learning model for identification, and judging whether the current driving state is a standard driving behavior or an illegal driving behavior according to the identification result.
在其中的一实施例，所述深度学习模型包括标准数据和违规数据，该标准数据对应于所述驾驶员的标准驾驶行为，该违规数据对应于所述驾驶员的接电话、玩手机或双手长时间脱离操作杆的违规驾驶行为。In one embodiment, the deep learning model includes standard data and violation data, where the standard data corresponds to the driver's standard driving behavior, and the violation data corresponds to the driver's answering a phone, playing with a mobile phone, or using his hands Driving violations involving prolonged disengagement of the operating lever.
在其中的一实施例，于步骤c)之后，该方法还包括：根据所述视频帧的帧顺序，依次输入单帧骨骼特征向量至所述深度学习模型进行训练，并更新所述深度学习模型。In one embodiment, after step c), the method further includes: sequentially inputting single-frame skeleton feature vectors to the deep learning model for training according to the frame sequence of the video frames, and updating the deep learning model .
在其中的一实施例，所述驾驶员的骨架数据为所述Kinect摄像机定位到驾驶员全身的25个关节点以及相应的关节点坐标值。In one embodiment, the skeleton data of the driver is the 25 joint points located on the driver's whole body by the Kinect camera and the corresponding joint point coordinate values.
在其中的一实施例，对于仅有坐姿的驾驶员，所述单帧骨骼特征向量由稳定不变的若干关节点坐标以及这些关节点之间所构成的夹角组成。In one embodiment, for a driver who only has a sitting posture, the single-frame skeleton feature vector is composed of a number of joint point coordinates that are stable and constant and the included angle formed by these joint points.
在其中的一实施例，于步骤d)之后，该方法还包括：延迟一预设的时间间隔，返回上述步骤a)，并循环执行上述步骤a)至上述步骤d)的驾驶行为识别过程。In one embodiment, after step d), the method further includes: delaying a preset time interval, returning to step a), and cyclically executing the driving behavior identification process from step a) to step d).
在其中的一实施例，上述步骤d)还包括：若识别结果为违规驾驶行为，则发出声光报警或者标志指示信号。In one embodiment, the above step d) further includes: if the identification result is illegal driving behavior, sending out an audible and visual alarm or a sign indicating signal.
采用本发明的基于Kinect摄像机的驾驶行为识别方法，首先利用Kinect摄像机采集的视频帧对前景目标驾驶员进行骨架追踪，并定位驾驶员的骨架数据；然后根据驾驶员的骨架数据，获取视频帧的每一帧中的驾驶员的多个关节点坐标值并对其进行归一化处理；接着对关节点坐标值进行特征提取，生成多个单帧骨骼特征向量；最后将多个单帧骨骼特征向量送入预设的深度学习模型进行识别，根据识别结果判断当前驾驶状态是标准驾驶行为或违规驾驶行为。相比于现有技术，本发明可根据驾驶员骨架的多个关节点坐标以及相应的骨骼特征向量来实现驾驶员操作行为的识别，达到报警或预警的防范效果，并且基于Kinect摄像机对骨骼信息提取时，不受光线的影响且精度更高、更可靠。By adopting the method for recognizing driving behavior based on the Kinect camera of the present invention, firstly, the skeleton of the foreground target driver is tracked by using the video frames collected by the Kinect camera, and the skeleton data of the driver is located; then, according to the skeleton data of the driver, the The multiple joint point coordinate values of the driver in each frame are normalized; then feature extraction is performed on the joint point coordinate values to generate multiple single-frame skeleton feature vectors; The vector is sent to the preset deep learning model for recognition, and the current driving state is judged whether the current driving behavior is standard driving behavior or illegal driving behavior according to the recognition result. Compared with the prior art, the present invention can realize the identification of the driver's operation behavior according to the coordinates of multiple joint points of the driver's skeleton and the corresponding skeleton feature vector, so as to achieve the prevention effect of alarm or early warning, and based on the Kinect camera, the skeleton information can be detected. When extracting, it is not affected by light and is more accurate and reliable.
附图说明Description of drawings
读者在参照附图阅读了本发明的具体实施方式以后，将会更清楚地了解本发明的各个方面。其中，Various aspects of the present invention will be more clearly understood by the reader after reading the detailed description of the invention with reference to the accompanying drawings. in,
图1示出本发明的基于Kinect摄像机的驾驶行为识别方法的流程框图。FIG. 1 shows a flowchart of the method for recognizing driving behavior based on the Kinect camera of the present invention.
图2示出图1的驾驶行为识别方法中使用的Kinect摄像机所提取的人体25个关节点的示意图。FIG. 2 shows a schematic diagram of 25 joint points of the human body extracted by the Kinect camera used in the driving behavior recognition method of FIG. 1 .
具体实施方式Detailed ways
为了使本申请所揭示的技术内容更加详尽与完备，可参照附图以及本发明的下述各种具体实施例，附图中相同的标记代表相同或相似的组件。然而，本领域的普通技术人员应当理解，下文中所提供的实施例并非用来限制本发明所涵盖的范围。此外，附图仅仅用于示意性地加以说明，并未依照其原尺寸进行绘制。In order to make the technical content disclosed in this application more detailed and complete, reference may be made to the accompanying drawings and the following various specific embodiments of the present invention, wherein the same symbols in the accompanying drawings represent the same or similar components. However, those of ordinary skill in the art should understand that the embodiments provided below are not intended to limit the scope covered by the present invention. Furthermore, the drawings are for schematic illustration only and are not drawn to their full scale.
下面参照附图，对本发明各个方面的具体实施方式作进一步的详细描述。The specific embodiments of various aspects of the present invention will be described in further detail below with reference to the accompanying drawings.
图1示出本发明的基于Kinect摄像机的驾驶行为识别方法的流程框图。图2示出图1的驾驶行为识别方法中使用的Kinect摄像机所提取的人体25个关节点的示意图。FIG. 1 shows a flowchart of the method for recognizing driving behavior based on the Kinect camera of the present invention. FIG. 2 shows a schematic diagram of 25 joint points of the human body extracted by the Kinect camera used in the driving behavior recognition method of FIG. 1 .
参照图1和图2，在该实施例中，本发明的基于Kinect摄像机的驾驶行为识别方法包括步骤S101～S107。相比于现有技术，本发明藉由上述步骤S101、步骤S103、步骤S105和步骤S107，根据驾驶员骨架的多个关节点坐标以及相应的骨骼特征向量来实现驾驶员操作行为的识别，达到报警或预警的防范效果，并且基于Kinect摄像机对骨骼信息提取时，不受光线的影响且精度更高、更可靠。Referring to FIG. 1 and FIG. 2 , in this embodiment, the method for recognizing driving behavior based on the Kinect camera of the present invention includes steps S101 to S107 . Compared with the prior art, the present invention realizes the identification of the driver's operation behavior according to the above-mentioned steps S101, S103, S105 and S107 according to the coordinates of multiple joint points of the driver's skeleton and the corresponding skeleton feature vectors, so as to achieve The prevention effect of alarm or early warning, and when extracting bone information based on the Kinect camera, it is not affected by light and has higher precision and reliability.
在步骤S101中，利用Kinect摄像机采集的视频帧对前景目标驾驶员进行骨架追踪，并定位驾驶员的骨架数据。较佳地，驾驶员的骨架数据为Kinect摄像机定位到驾驶员全身的25个关节点以及相应的关节点坐标值。如图2所示，骨架的25个关节点的位置及含义分别为：In step S101, the skeleton of the foreground target driver is tracked by using the video frames collected by the Kinect camera, and the skeleton data of the driver is located. Preferably, the skeleton data of the driver is the 25 joint points located on the driver's whole body by the Kinect camera and the corresponding joint point coordinate values. As shown in Figure 2, the positions and meanings of the 25 joint points of the skeleton are:
Head-头部，Neck-颈部，SpineShoulder-脊柱肩膀，SpineMid-脊柱中心，SpineBase-脊柱底部，ShoulderLeft-左肩，ElbowLeft-左肘，WristLeft-左腕，HandLeft-左手，HandTipLeft-左指尖，ThumbLeft-左拇指，HipLeft-左臀，ShoulderRight-右肩，ElbowRight-右肘，WristRight-右腕，HandRight-右手，HandTipRight-右指尖，ThumbRight-右拇指，HipRight-右臀，KneeLeft-左膝，AnkleLeft-左脚踝，FootLeft-左脚，KneeRight-右膝，AnkleRight-右脚踝，FootRight-右脚。Head-Head, Neck-Neck, SpineShoulder-Spine Shoulder, SpineMid-Spine Center, SpineBase-Spine Bottom, ShoulderLeft-Left Shoulder, ElbowLeft-Left Elbow, WristLeft-Left Wrist, HandLeft-Left Hand, HandTipLeft-Left Fingertip, ThumbLeft- Left Thumb, HipLeft - Left Hip, ShoulderRight - Right Shoulder, ElbowRight - Right Elbow, WristRight - Right Wrist, HandRight - Right Hand, HandTipRight - Right Fingertip, ThumbRight - Right Thumb, HipRight - Right Hip, KneeLeft - Left Knee, AnkleLeft - Left Ankle, FootLeft - Left Foot, KneeRight - Right Knee, AnkleRight - Right Ankle, FootRight - Right Foot.
在步骤S103中，根据驾驶员的骨架数据，获取视频帧的每一帧中的驾驶员的多个关节点坐标值并对其进行归一化处理，其中关节点坐标值之间的内部关系用于表征所述驾驶员的驾驶行为。In step S103, according to the skeleton data of the driver, multiple joint point coordinate values of the driver in each frame of the video frame are obtained and normalized, wherein the internal relationship between the joint point coordinate values is determined by to characterize the driver's driving behavior.
在步骤S105中，对关节点坐标值进行特征提取，生成多个单帧骨骼特征向量。在一些实施例中，对于仅有坐姿的驾驶员，单帧骨骼特征向量由稳定不变的若干关节点坐标以及这些关节点之间所构成的夹角组成。例如，由于驾驶员的驾驶行为均为坐姿，则主要提取骨架轮廓的上半身的关节信息，诸如SpineBase(脊柱底部)、Head(头部)、ElbowLeft(左肘)、WristLeft(左腕)、ElbowRight(右肘)、WristRight(右腕)几个关节点的坐标。In step S105, feature extraction is performed on the joint point coordinate values to generate multiple single-frame skeleton feature vectors. In some embodiments, for a driver who only has a sitting posture, a single-frame skeleton feature vector consists of several joint point coordinates that are stable and constant and the included angles formed between these joint points. For example, since the driving behavior of the driver is all sitting posture, the joint information of the upper body of the skeleton outline is mainly extracted, such as SpineBase (bottom of spine), Head (head), ElbowLeft (left elbow), WristLeft (left wrist), ElbowRight (right). Elbow), WristRight (right wrist) coordinates of several joint points.
在步骤S107中，将多个单帧骨骼特征向量送入预设的深度学习模型进行识别，根据识别结果判断当前驾驶状态是标准驾驶行为或违规驾驶行为。较佳地，该深度学习模型包括标准数据和违规数据，该标准数据对应于驾驶员的标准驾驶行为，该违规数据对应于驾驶员的接电话、玩手机或双手长时间脱离操作杆的违规驾驶行为。In step S107, a plurality of single-frame skeleton feature vectors are sent to a preset deep learning model for identification, and according to the identification result, it is determined whether the current driving state is a standard driving behavior or an illegal driving behavior. Preferably, the deep learning model includes standard data and violation data, the standard data corresponds to the driver's standard driving behavior, and the violation data corresponds to the driver's illegal driving of answering the phone, playing with a mobile phone, or taking his hands off the joystick for a long time. Behavior.
依据一些实施例，在上述步骤S105之后，该驾驶行为识别方法还包括根据视频帧的帧顺序，依次输入单帧骨骼特征向量至深度学习模型进行训练，并更新深度学习模型。According to some embodiments, after the above step S105, the driving behavior recognition method further includes sequentially inputting single-frame skeleton feature vectors to the deep learning model for training according to the frame sequence of the video frames, and updating the deep learning model.
依据一些实施例，在上述步骤S107之后，该驾驶行为识别方法还包括延迟一预设的时间间隔，返回上述步骤S101，并循环执行上述步骤S101至上述步骤S107的驾驶行为识别过程。此外，在步骤S107中，若识别结果为违规驾驶行为，则发出声光报警或者标志指示信号。According to some embodiments, after the above step S107, the driving behavior recognition method further includes delaying a preset time interval, returning to the above step S101, and cyclically executing the driving behavior recognition process from the above step S101 to the above step S107. In addition, in step S107, if the identification result is illegal driving behavior, an audible and visual alarm or a sign indicating signal is issued.
为检验本发明的驾驶行为识别方法的准确率，发明人采用Kinect摄像机采集视频，共包括20帧，共3200个驾驶行为序列，其中，随机抽取2400个驾驶行为序列建立深度学习模型，余下的800个驾驶行为序列进行测试。仿真实验数据表明，标准姿势的识别准确率为100％，玩手机的违规行为的识别准确率为98.0％，接电话的违规行为的识别准确率为98.0％，双手长时间脱离操作杆的违规行为的识别准确率为96.0％。In order to test the accuracy of the driving behavior recognition method of the present invention, the inventor uses a Kinect camera to collect videos, including 20 frames in total, and a total of 3200 driving behavior sequences, of which 2400 driving behavior sequences are randomly selected to establish a deep learning model, and the remaining 800 test driving behavior sequences. The simulation experimental data shows that the recognition accuracy of standard poses is 100%, the recognition accuracy of illegal behaviors of playing mobile phones is 98.0%, the recognition accuracy of illegal behaviors of answering the phone is 98.0%, and the violations of violations involving hands off the operating lever for a long time The recognition accuracy is 96.0%.
采用本发明的基于Kinect摄像机的驾驶行为识别方法，首先利用Kinect摄像机采集的视频帧对前景目标驾驶员进行骨架追踪，并定位驾驶员的骨架数据；然后根据驾驶员的骨架数据，获取视频帧的每一帧中的驾驶员的多个关节点坐标值并对其进行归一化处理；接着对关节点坐标值进行特征提取，生成多个单帧骨骼特征向量；最后将多个单帧骨骼特征向量送入预设的深度学习模型进行识别，根据识别结果判断当前驾驶状态是标准驾驶行为或违规驾驶行为。相比于现有技术，本发明可根据驾驶员骨架的多个关节点坐标以及相应的骨骼特征向量来实现驾驶员操作行为的识别，达到报警或预警的防范效果，并且基于Kinect摄像机对骨骼信息提取时，不受光线的影响且精度更高、更可靠。By adopting the method for recognizing driving behavior based on the Kinect camera of the present invention, firstly, the skeleton of the foreground target driver is tracked by using the video frames collected by the Kinect camera, and the skeleton data of the driver is located; then, according to the skeleton data of the driver, the The multiple joint point coordinate values of the driver in each frame are normalized; then feature extraction is performed on the joint point coordinate values to generate multiple single-frame skeleton feature vectors; The vector is sent to the preset deep learning model for recognition, and the current driving state is judged whether the current driving behavior is standard driving behavior or illegal driving behavior according to the recognition result. Compared with the prior art, the present invention can realize the identification of the driver's operation behavior according to the coordinates of multiple joint points of the driver's skeleton and the corresponding skeleton feature vector, so as to achieve the prevention effect of alarm or early warning, and based on the Kinect camera, the skeleton information can be detected. When extracting, it is not affected by light and is more accurate and reliable.
上文中，参照附图描述了本发明的具体实施方式。但是，本领域中的普通技术人员能够理解，在不偏离本发明的精神和范围的情况下，还可以对本发明的具体实施方式作各种变更和替换。这些变更和替换都落在本发明权利要求书所限定的范围内。Hereinabove, specific embodiments of the present invention have been described with reference to the accompanying drawings. However, those skilled in the art can understand that various changes and substitutions can be made to the specific embodiments of the present invention without departing from the spirit and scope of the present invention. These modifications and substitutions fall within the scope defined by the claims of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910088268.8ACN109886150A (en) | 2019-01-29 | 2019-01-29 | A driving behavior recognition method based on Kinect camera |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910088268.8ACN109886150A (en) | 2019-01-29 | 2019-01-29 | A driving behavior recognition method based on Kinect camera |
| Publication Number | Publication Date |
|---|---|
| CN109886150Atrue CN109886150A (en) | 2019-06-14 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910088268.8APendingCN109886150A (en) | 2019-01-29 | 2019-01-29 | A driving behavior recognition method based on Kinect camera |
| Country | Link |
|---|---|
| CN (1) | CN109886150A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110490143A (en)* | 2019-08-21 | 2019-11-22 | 西安工程大学 | A kind of action identification method of adaptive model |
| CN112046662A (en)* | 2020-08-13 | 2020-12-08 | 哈尔滨工业大学（深圳） | Walking-replacing following robot and walking-replacing following method thereof |
| CN112307846A (en)* | 2019-08-01 | 2021-02-02 | 北京新联铁集团股份有限公司 | Analysis method for violation of crew service |
| CN112446352A (en)* | 2020-12-14 | 2021-03-05 | 深圳地平线机器人科技有限公司 | Behavior recognition method, behavior recognition device, behavior recognition medium, and electronic device |
| CN114005102A (en)* | 2020-07-28 | 2022-02-01 | 株洲中车时代电气股份有限公司 | Driver's illegal driving behavior detection method, device, device and readable storage medium |
| CN114694127A (en)* | 2022-04-20 | 2022-07-01 | 南京邮电大学 | Driving behavior recognition method and system based on depth image |
| WO2022142786A1 (en)* | 2020-12-30 | 2022-07-07 | 中兴通讯股份有限公司 | Driving behavior recognition method, and device and storage medium |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105469073A (en)* | 2015-12-16 | 2016-04-06 | 安徽创世科技有限公司 | Kinect-based call making and answering monitoring method of driver |
| CN107657236A (en)* | 2017-09-29 | 2018-02-02 | 厦门知晓物联技术服务有限公司 | Vehicle security drive method for early warning and vehicle-mounted early warning system |
| CN108446678A (en)* | 2018-05-07 | 2018-08-24 | 同济大学 | A kind of dangerous driving behavior recognition methods based on skeleton character |
| CN108830215A (en)* | 2018-06-14 | 2018-11-16 | 南京理工大学 | Hazardous act recognition methods based on personnel's framework information |
| CN108921044A (en)* | 2018-06-11 | 2018-11-30 | 大连大学 | Driver's decision feature extracting method based on depth convolutional neural networks |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105469073A (en)* | 2015-12-16 | 2016-04-06 | 安徽创世科技有限公司 | Kinect-based call making and answering monitoring method of driver |
| CN107657236A (en)* | 2017-09-29 | 2018-02-02 | 厦门知晓物联技术服务有限公司 | Vehicle security drive method for early warning and vehicle-mounted early warning system |
| CN108446678A (en)* | 2018-05-07 | 2018-08-24 | 同济大学 | A kind of dangerous driving behavior recognition methods based on skeleton character |
| CN108921044A (en)* | 2018-06-11 | 2018-11-30 | 大连大学 | Driver's decision feature extracting method based on depth convolutional neural networks |
| CN108830215A (en)* | 2018-06-14 | 2018-11-16 | 南京理工大学 | Hazardous act recognition methods based on personnel's framework information |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112307846A (en)* | 2019-08-01 | 2021-02-02 | 北京新联铁集团股份有限公司 | Analysis method for violation of crew service |
| CN110490143A (en)* | 2019-08-21 | 2019-11-22 | 西安工程大学 | A kind of action identification method of adaptive model |
| CN114005102A (en)* | 2020-07-28 | 2022-02-01 | 株洲中车时代电气股份有限公司 | Driver's illegal driving behavior detection method, device, device and readable storage medium |
| CN112046662A (en)* | 2020-08-13 | 2020-12-08 | 哈尔滨工业大学（深圳） | Walking-replacing following robot and walking-replacing following method thereof |
| CN112046662B (en)* | 2020-08-13 | 2023-01-17 | 哈尔滨工业大学（深圳） | A walking and following robot and its walking and following method |
| CN112446352A (en)* | 2020-12-14 | 2021-03-05 | 深圳地平线机器人科技有限公司 | Behavior recognition method, behavior recognition device, behavior recognition medium, and electronic device |
| WO2022142786A1 (en)* | 2020-12-30 | 2022-07-07 | 中兴通讯股份有限公司 | Driving behavior recognition method, and device and storage medium |
| CN114764912A (en)* | 2020-12-30 | 2022-07-19 | 中兴通讯股份有限公司 | Driving behavior recognition method, device and storage medium |
| CN114694127A (en)* | 2022-04-20 | 2022-07-01 | 南京邮电大学 | Driving behavior recognition method and system based on depth image |
| Publication | Publication Date | Title |
|---|---|---|
| CN109886150A (en) | A driving behavior recognition method based on Kinect camera | |
| JP4860749B2 (en) | Apparatus, system, and method for determining compatibility with positioning instruction in person in image | |
| Ahmed et al. | Vision based hand gesture recognition using dynamic time warping for Indian sign language | |
| US8611587B2 (en) | Device, system and method for determining compliance with an instruction by a figure in an image | |
| WO2015165365A1 (en) | Facial recognition method and system | |
| CN108304819B (en) | Gesture recognition system and method, and storage medium | |
| TW202201275A (en) | Device and method for scoring hand work motion and storage medium | |
| CN112101315A (en) | Deep learning-based exercise judgment guidance method and system | |
| CN107909020A (en) | A kind of micro- expression stage of development detection method of light stream vectors based on wave filter design | |
| WO2017161734A1 (en) | Correction of human body movements via television and motion-sensing accessory and system | |
| CN116259002A (en) | A video-based human risk behavior analysis method | |
| CN114758294A (en) | Method, device, equipment and storage medium for monitoring dangerous behaviors of distribution center | |
| CN102622596B (en) | Interactive localization method towards living things feature recognition | |
| CN113989919A (en) | A gait capture recognition method for video surveillance and its application | |
| KR101745601B1 (en) | method for finger counting by using image processing and apparatus adopting the method | |
| CN110889326B (en) | Human body detection-based queue-jumping behavior monitoring and warning system, method, device and storage medium | |
| CN107767366B (en) | A transmission line fitting method and device | |
| WO2015131710A1 (en) | Method and device for positioning human eyes | |
| CN118172704A (en) | Sedentary reminding method based on machine vision and electronic equipment | |
| CN114529991B (en) | Pedestrian attribute identification method and device, storage medium and electronic device | |
| CN104102895A (en) | Equipment and method used for detecting body parts of photographed user | |
| CN114550287A (en) | Human body key point-based method for detecting abnormal behavior of personnel in transformer substation scene | |
| CN119919690B (en) | Experimental compliance detection method and device based on artificial intelligence | |
| CN113128375A (en) | Image recognition method, electronic device and computer-readable storage medium | |
| CN116824424B (en) | A campus protection automatic alarm method based on deep learning |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | Application publication date:20190614 | |
| WD01 | Invention patent application deemed withdrawn after publication |