CN115861926A - Passenger behavior identification method and system in car type elevator and electronic equipment - Google Patents
Passenger behavior identification method and system in car type elevator and electronic equipment Download PDFInfo
- Publication number
- CN115861926A CN115861926A CN202211512182.1A CN202211512182A CN115861926A CN 115861926 A CN115861926 A CN 115861926A CN 202211512182 A CN202211512182 A CN 202211512182A CN 115861926 A CN115861926 A CN 115861926A
- Authority
- CN
- China
- Prior art keywords
- passenger
- video
- neural network
- slowfast
- behavior
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a passenger behavior recognition method, a passenger behavior recognition system and electronic equipment in a car elevator, relates to the technical field of behavior recognition, and aims to obtain a monitoring video in the car elevator; intercepting a passenger video segment in a monitoring video by using a target detection model and a target tracking algorithm; inputting the passenger video segment into a behavior recognition model to obtain the behavior category of a passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution. According to the invention, by constructing the target detection model and the behavior recognition model, dangerous behaviors of passengers in the car type elevator can be accurately recognized, and thus the running safety of the elevator is improved.
Description
Technical Field
The invention relates to the technical field of behavior recognition, in particular to a method and a system for recognizing passenger behaviors in a car elevator and electronic equipment.
Background
With the heavy use of car elevators, some safety issues of elevators are of increasing concern. In recent years, reports about safety accidents of elevators caused by dangerous behaviors (such as opening of elevator doors) of passengers in elevator cars occur, and the behaviors can affect safe operation of the elevators on one hand and possibly threaten personal safety of the passengers on the other hand.
Although most of the existing car elevators are provided with monitoring cameras, the monitoring cameras are generally used for detecting the flow of the elevators. Aiming at the problem of identifying dangerous behaviors of passengers in an elevator, only a few elevators are provided with voice broadcasting modules, workers are arranged to be on duty in a monitoring room, the elevators are reminded through the voice broadcasting modules according to the real-time situation in elevator cars and are connected for rescue after accidents occur, however, places with large passenger flow such as markets and communities are provided with a plurality of car elevators, the frequency of accidents is high, monitoring videos of the plurality of car elevators in the monitoring room are usually only arranged to be checked by one to two on-duty security personnel, the on-duty security personnel cannot give consideration to the dangerous behaviors made by the passengers in the plurality of elevators for a long time and remind the passengers, and meanwhile, even the accidents cannot be screened and responded, the dangerous behaviors cannot be reminded, so that the dangerous behaviors cannot be timely accumulated for a long time to accelerate aging of the elevators and the increase of failure rate, sudden elevator failure rate is caused, commercial operation is influenced, and economic loss is directly or indirectly caused.
Disclosure of Invention
The invention aims to provide a method, a system and electronic equipment for identifying passenger behaviors in a car elevator, which can accurately identify dangerous behaviors of passengers in the car elevator and further improve the running safety of the elevator.
In order to achieve the purpose, the invention provides the following scheme:
a passenger behavior identification method in a car type elevator comprises the following steps:
acquiring a monitoring video in the car type elevator;
intercepting a passenger video segment in the monitoring video by using a target detection model and a target tracking algorithm; the target detection model is obtained by training a YOLOv5 neural network according to the marked historical monitoring videos of the multiple car elevators;
inputting the passenger video segment into a behavior recognition model to obtain the behavior category of the passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution.
Optionally, after the step of inputting the passenger video segment into the behavior recognition model to obtain the behavior category of the passenger in the passenger video segment, the method further includes:
and determining whether to control the corresponding car type elevator to send out an alarm signal according to the behavior category of the passenger.
Optionally, the intercepting a passenger video segment in the surveillance video by using a target detection model and a target tracking algorithm includes:
performing framing processing on the monitoring video to obtain a monitoring video frame sequence;
inputting the monitoring video frame sequence into a target detection model, and identifying passengers with the monitoring video;
and tracking each passenger appearing in the monitoring video respectively by using a target tracking algorithm, and determining a plurality of continuous video frames corresponding to the passenger with the appearing frame number greater than the frame number threshold value as passenger video segments.
Optionally, the target tracking algorithm is a deepsort algorithm.
Optionally, before obtaining the monitoring video in the car elevator, the method further includes:
acquiring historical monitoring videos of a plurality of car type elevators;
framing the plurality of historical monitoring videos to obtain a plurality of historical monitoring video frames;
marking passengers in the plurality of historical monitoring video frames respectively to obtain a plurality of marked historical monitoring video frames;
and training the YOLOv5 neural network by taking a plurality of labeled historical monitoring video frames as input and passengers labeled in the labeled historical monitoring video frames as output to obtain the target detection model.
Optionally, before the obtaining of the monitoring video in the car elevator, the method further includes:
replacing the convolution in the SlowFast neural network with non-local convolution to obtain an improved SlowFast neural network;
acquiring a plurality of historical passenger video segments;
labeling the passenger behavior category in each historical passenger video segment;
and training the improved SlowFast neural network by taking the historical passenger video band as input and the passenger behavior category as output to obtain a behavior recognition model.
A system for passenger behavior recognition in a car elevator, comprising:
the monitoring video acquisition module is used for acquiring monitoring videos in the car type elevator;
the target detection module is used for intercepting a passenger video segment in the monitoring video by utilizing a target detection model and a target tracking algorithm; the target detection model is obtained by training a YOLOv5 neural network according to the marked historical monitoring videos of the multiple car elevators;
the behavior type determining module is used for inputting the passenger video segment into a behavior recognition model to obtain the behavior type of the passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video bands; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution.
An electronic device comprising a memory for storing a computer program and a processor running the computer program to cause the electronic device to perform the method of passenger behavior identification in a car elevator.
Optionally, the memory is a readable storage medium.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a passenger behavior identification method, a passenger behavior identification system and electronic equipment in a car elevator, which are used for acquiring a monitoring video in the car elevator; intercepting a passenger video segment in a monitoring video by using a target detection model and a target tracking algorithm; the target detection model is obtained by training a YOLOv5 neural network according to the marked historical monitoring videos of the multiple car elevators; inputting the passenger video segment into a behavior recognition model to obtain the behavior category of a passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution. According to the invention, by constructing the target detection model and the behavior recognition model, dangerous behaviors of passengers in the car type elevator can be accurately recognized, and thus the running safety of the elevator is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a flowchart of a passenger behavior recognition method in a car elevator according to embodiment 1 of the present invention;
FIG. 2 is a flowchart of behavior recognition model training in embodiment 1 of the present invention;
FIG. 3 is a flowchart of a method for using a behavior recognition model according to embodiment 1 of the present invention;
FIG. 4 is a flowchart of the non-local convolution calculation in embodiment 1 of the present invention;
fig. 5 is a flow chart of the improved calculation of the SlowFast neural network in embodiment 1 of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method, a system and electronic equipment for identifying passenger behaviors in a car elevator, which can accurately identify dangerous behaviors of passengers in the car elevator and further improve the running safety of the elevator.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Example 1
As shown in fig. 1 and fig. 3, the present embodiment provides a method for identifying passenger behavior in a car elevator, including:
step 101: and acquiring a monitoring video in the car type elevator.
Step 102: intercepting a passenger video segment in a monitoring video by using a target detection model and a target tracking algorithm; the target detection model is obtained by training a Yolov5 neural network according to the marked historical monitoring videos of the multiple car elevators. Wherein, the target tracking algorithm is a deepsort algorithm.
For example, step 102 includes:
step 1021: and performing framing processing on the monitoring video to obtain a monitoring video frame sequence.
Step 1022: and inputting the monitoring video frame sequence into a target detection model, and identifying passengers with the monitoring video.
Step 1023: and tracking each passenger appearing in the monitoring video respectively by using a target tracking algorithm, and determining a plurality of continuous video frames corresponding to the passenger with the appearing frame number greater than the frame number threshold value as passenger video segments.
Specifically, video segments including passengers shot by a monitoring camera located in the car elevator are acquired, and a training set and a verification set are generated by using the video segments. All passengers in the image are selected by using a YOLOv5 network frame for each frame image of the video band and tracked by using a depersort algorithm, for example, the passenger A is tracked, and the video band from the action starting frame to the action ending frame in the video is intercepted, and the category of the video band is marked as the behavior category of the passenger A. All samples were then run as 9: the scale of 1 is divided into the original training set and the original validation set.
Step 103: inputting the passenger video segment into a behavior recognition model to obtain the behavior category of a passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution; the improved SlowFast neural network is shown in fig. 4 and 5.
Step 104: and determining whether to control the corresponding car type elevator to send out an alarm signal or not according to the behavior category of the passenger.
As shown in fig. 2, the method for identifying a passenger behavior in a car elevator provided in this embodiment further includes, before step 101: step 105-step 108.
Step 105: acquiring historical monitoring videos of a plurality of car type elevators;
step 106: framing a plurality of historical monitoring videos to obtain a plurality of historical monitoring video frames;
step 107: marking passengers in the plurality of historical monitoring video frames respectively to obtain a plurality of marked historical monitoring video frames;
step 108: and training the YOLOv5 neural network by taking a plurality of marked historical monitoring video frames as input and passengers marked in the marked historical monitoring video frames as output to obtain a target detection model.
In addition, the method for identifying passenger behavior in a car elevator provided in this embodiment further includes, before step 101: step 109-step 1012.
Step 109: and replacing the convolution in the SlowFast neural network with non-local convolution to obtain the improved SlowFast neural network.
Step 1010: a plurality of historical passenger video segments are acquired.
Step 1011: and marking the passenger behavior category in each historical passenger video segment respectively.
Step 1012: and training the improved SlowFast neural network by taking the historical passenger video band as input and the passenger behavior category as output to obtain a behavior recognition model.
Because the training set is limited in scale, more realistic scenes can be simulated by performing data amplification on the basis of the existing training set, and the richness of data and the generalization of a final training model can be effectively improved. The data amplification method is characterized in that the following data amplification methods are added according to the special situation of the car elevator besides the expansion method of the SlowFast network:
(1) Noise addition: gaussian noise is added to the video in the training set, and the Gaussian noise refers to the noise of which the probability density function follows Gaussian distribution. A noise is said to be white Gaussian if its amplitude distribution follows a Gaussian distribution and its power spectral density is uniformly distributed. Adding salt and pepper noise to the video in the training set, wherein the salt and pepper noise refers to two kinds of noise, one kind of noise is salt noise, and the other kind of noise is pepper noise. Salt = white (0) and pepper = black (255). The former is high gray noise, and the latter is low gray noise. Two kinds of noise generally appear at the same time, and appear on the image as black and white noise spots.
(2) Histogram equalization: histogram equalization is a simple and effective image enhancement technique, which changes the gray scale of each pixel in an image by changing the histogram of the image, and is mainly used for enhancing the contrast of the image with a small dynamic range. The original image may be concentrated in a narrow interval due to its gray distribution, resulting in an insufficiently sharp image. For example, an overexposed image will have its gray levels centered in the high brightness range, while an underexposure will have its gray levels centered in the low brightness range. By adopting histogram equalization, the histogram of the original image can be converted into a form of uniform distribution (equalization), so that the dynamic range of gray value difference between pixels is increased, and the effect of enhancing the overall contrast of the image is achieved. In other words, the basic principle of histogram equalization is: the gray values with a large number of pixels in the image (namely the gray values which play a main role in the picture) are widened, and the gray values with a small number of pixels (namely the gray values which do not play a main role in the picture) are merged, so that the contrast is increased, the image is clear, and the aim of enhancement is fulfilled.
(3) Mosaic operation: because the position of the human body in the elevator can not change greatly, the Mosaic operation is provided, the idea is to intercept passengers in different videos by using a YOLOv5 network, for example, intercept passengers A, B, C and D, and then splice each frame of the passengers in the videos into one frame as training data, so that the advantage of enriching the background of the videos is realized, and the robustness of the network at the training position is improved.
Using training set after data amplificationTraining a proposed improved SlowFast network, the specific improvement of which is as follows: the improved SlowFast sparsely samples a series of short segments from the entire video, each of which will give its own preliminary prediction of behavior class, and the video-level prediction results are derived from the "consensus" of these segments. In the learning process, the loss value of the video level prediction is optimized by iteratively updating the model parameters. A given video is first divided equally and then subjected to subsequent processing. Assuming that a video is equally divided into K segments, and the sampling frame number of a slow channel of the SlowFast network is T, the time length of the slow channel of the improved SlowFast network is T/K, and the rest conditions of the original SlowFast network are kept unchanged. The class scores of the different segments are fused using a segment consensus function to produce a segment consensus, which is a video-level prediction. The prediction fusion of all modes then yields the final prediction result. Specifically, given a video V, it is divided into K segments { S } at equal intervals 1, S 2, ...S K }. The improved SlowFast then models a series of fragments as follows:
slowfast((S 1 ,T 1 ),(S 2, T 2 ),...(S K, T K ))＝H(G(F(S 1, T 1 ；W),F(S 2, T 2 ；W),...F(S K, T K ；W)))。
wherein T is K Is a slave S K Of the randomly extracted set of T frame images. F (S) K ,T K (ii) a W) function represents the convolution network effect (S) using W as a parameter K, T K ) Video group, function Return (S) K, T K ) Scores against all categories. The segment consensus function G combines the class score outputs of the plurality of short segments to obtain consensus among them on the class hypothesis. The probability that the whole video segment belongs to each behavior class is predicted based on the consensus prediction function H. Wherein (S) K, T K ) A video set is composed and input into the SlowFast network. W is a parameter of the SlowFast network after training. F (S) K ,T K (ii) a W) function represents the convolution network effect (S) using W as a parameter K ,T K ) Video group, function F returns (S) K ,T K ) Scores against all categories. G is a segment consensus function, whose effect is that the segment consensus function G combines the class score outputs of multiple short segments to obtain a consensus score between them for a class hypothesis. The probability that the whole video segment belongs to each behavior class is predicted based on the consensus prediction function H.
The normal convolution is replaced by a non-partial convolution. In the conventional Convolutional Neural Network (CNN) and fully-connected neural network (DNN), the computation of convolutional layers is only weighted summation of surrounding features, and generally, the computation of current layers only depends on the result of the previous layer, and most networks currently use convolutional kernels with the size of 1 × 1 or 3 × 3, and extraction of long-distance related features is insufficient. Both hole convolution and deformable convolution increase the receptive field, but the magnitude of the increase is limited. Non-local convolution was originally used to handle the dependency between capturing distant pixels in the video image classification task. The traditional network generally increases the receptive field by repeating the convolution operation for many times to capture the dependency relationship between the remote pixels, but has the problems of low calculation efficiency, difficult optimization and the like. The non-local convolution represents the response at one location as a weighted average of the features at various locations in the input feature map. The non-local operation aggregates the input information according to the similarity of the input, and is defined as follows:
where i is the index of the output position time of the response to be calculated and j is the index of all possible positions. x (including x) i And x j ) Representing the input signal and y is the output signal of the same magnitude as x. The function f is the relation between the index i used to calculate the output position time of the response to be calculated and all possible associated positions j. g (.) represents the characteristic value at the position j where the input signal is calculated. C (.) represents the normalization parameter. The non-partial convolution is shown in fig. 3.
The improved SlowFast network is trained by using a training set after data amplification, an Adam optimization method and an initial learning rate of 0.001 are adopted for training in the training, a verification set is used for verifying the performance of the current model once every 10 epochs, and the model parameter with the highest accuracy of the verification set in the first 100 epochs is selected as a prediction model.
And finally, performing behavior classification on the video to be detected by using a trained improved SlowFast network: firstly, a video segment containing passengers shot by a monitoring camera positioned in a car elevator is obtained. And selecting all passengers in the image by using a YOLOv5 network frame for each frame image of the video band, tracking by using a depsort algorithm, such as tracking a passenger A, tracking the passenger A in the video, and inputting the video of the passenger A into a trained SlowFast network for classification after downsampling when the number of frames belonging to the passenger A reaches a set threshold value.
According to the embodiment, the target detection network and the behavior recognition network are combined, each passenger in the car elevator is recognized by the target detection network, each passenger is extracted from video data and input into the behavior recognition network, and dangerous behaviors of the passenger are screened out by the behavior recognition network. The behavior recognition network SlowFast network is improved, the SlowFast network has strong modeling capacity on short-term behaviors, but has poor modeling capacity on long-term behaviors, the advantages of the TSN network are utilized originally to segment the behavior videos of passengers, the segmented videos are respectively input into the SlowFast networks with shared weights, then the classification scores of the segmented behavior videos output by each SlowFast network are utilized, and the classification scores of the whole behavior videos are output by a consensus prediction function. Aiming at the characteristics that the light change of a car elevator is large, the monitoring noise of the elevator is large, and the relative change of the position of a passenger in the elevator is small, on the basis of a data expansion technology of a SlowFast network, a data expansion technology of adding Gaussian and salt-and-pepper noise and histogram equalization is introduced, and the precision of passenger behavior identification is improved.
Example 2
In order to implement the method corresponding to the above embodiment to achieve the corresponding functions and technical effects, the following provides a passenger behavior recognition system in a car elevator, including:
and the monitoring video acquisition module is used for acquiring monitoring videos in the car type elevator.
The target detection module is used for intercepting passenger video segments in the monitoring video by utilizing a target detection model and a target tracking algorithm; the target detection model is obtained by training a Yolov5 neural network according to the marked historical monitoring videos of the multiple car elevators.
The behavior type determining module is used for inputting the passenger video segment into the behavior recognition model to obtain the behavior type of the passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video bands; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution.
Example 3
The present embodiment provides an electronic device, which includes a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to make the electronic device execute the passenger behavior identification method in the car elevator in any embodiment 1. Wherein the memory is a readable storage medium.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (9)
1. A passenger behavior recognition method in a car type elevator is characterized by comprising the following steps:
acquiring a monitoring video in the car type elevator;
intercepting a passenger video segment in the monitoring video by using a target detection model and a target tracking algorithm; the target detection model is obtained by training a YOLOv5 neural network according to the marked historical monitoring videos of the multiple car elevators;
inputting the passenger video segment into a behavior recognition model to obtain the behavior category of the passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution.
2. The method according to claim 1, further comprising, after the step of inputting the passenger video segment into the behavior recognition model to obtain the behavior category of the passenger in the passenger video segment:
and determining whether to control the corresponding car type elevator to send out an alarm signal or not according to the behavior category of the passenger.
3. The method according to claim 1, wherein the intercepting a passenger video segment in the surveillance video by using a target detection model and a target tracking algorithm comprises:
performing framing processing on the monitoring video to obtain a monitoring video frame sequence;
inputting the monitoring video frame sequence into a target detection model, and identifying passengers with monitoring videos;
and respectively tracking each passenger appearing in the monitoring video by using a target tracking algorithm, and determining a plurality of continuous video frames corresponding to the passenger with the appearing frame number larger than the frame number threshold value as passenger video segments.
4. The method as claimed in claim 1, wherein the target tracking algorithm is a deepsort algorithm.
5. The method of claim 1, further comprising, prior to the obtaining the surveillance video in the car elevator:
acquiring historical monitoring videos of a plurality of car type elevators;
framing the plurality of historical monitoring videos to obtain a plurality of historical monitoring video frames;
marking passengers in the plurality of historical monitoring video frames respectively to obtain a plurality of marked historical monitoring video frames;
and training a YOLOv5 neural network by taking a plurality of marked historical monitoring video frames as input and passengers marked in the marked historical monitoring video frames as output to obtain the target detection model.
6. The method of claim 1, further comprising, prior to the obtaining the surveillance video in the car elevator:
replacing the convolution in the SlowFast neural network with non-local convolution to obtain an improved SlowFast neural network;
acquiring a plurality of historical passenger video segments;
labeling the passenger behavior category in each historical passenger video segment;
and training the improved SlowFast neural network by taking the historical passenger video band as input and the passenger behavior category as output to obtain a behavior recognition model.
7. A system for identifying passenger behavior in a car elevator, comprising:
the monitoring video acquisition module is used for acquiring monitoring videos in the car elevator;
the target detection module is used for intercepting a passenger video segment in the monitoring video by utilizing a target detection model and a target tracking algorithm; the target detection model is obtained by training a YOLOv5 neural network according to the marked historical monitoring videos of the multiple car elevators;
the behavior type determining module is used for inputting the passenger video segment into a behavior recognition model to obtain the behavior type of the passenger in the passenger video segment; the behavior recognition model is obtained by training an improved SlowFast neural network by utilizing a plurality of marked historical passenger video segments; the improved SlowFast neural network is obtained by replacing convolution in the SlowFast neural network with non-local convolution.
8. An electronic device, comprising a memory for storing a computer program and a processor for executing the computer program to cause the electronic device to execute a passenger behavior recognition method in a car elevator according to any one of claims 1 to 6.
9. An electronic device according to claim 8, wherein the memory is a readable storage medium.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211512182.1A CN115861926A (en) | 2022-11-29 | 2022-11-29 | Passenger behavior identification method and system in car type elevator and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211512182.1A CN115861926A (en) | 2022-11-29 | 2022-11-29 | Passenger behavior identification method and system in car type elevator and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115861926A true CN115861926A (en) | 2023-03-28 |
Family
ID=85667818
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211512182.1A Pending CN115861926A (en) | 2022-11-29 | 2022-11-29 | Passenger behavior identification method and system in car type elevator and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115861926A (en) |
-
2022
- 2022-11-29 CN CN202211512182.1A patent/CN115861926A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110458844B (en) | Semantic segmentation method for low-illumination scene | |
| CN108230291B (en) | Object recognition system training method, object recognition method, device and electronic equipment | |
| CN113449660B (en) | Abnormal event detection method of space-time variation self-coding network based on self-attention enhancement | |
| CN110516536A (en) | A kind of Weakly supervised video behavior detection method for activating figure complementary based on timing classification | |
| CN110633643A (en) | Abnormal behavior detection method and system for smart community | |
| CN109255326B (en) | Traffic scene smoke intelligent detection method based on multi-dimensional information feature fusion | |
| WO2022116616A1 (en) | Behavior recognition method based on conversion module | |
| CN110390308B (en) | Video behavior identification method based on space-time confrontation generation network | |
| CN117333753B (en) | Fire detection method based on PD-YOLO | |
| CN110059675A (en) | A kind of robot identifies road traffic law enforcement behavior and provides the method for standardization auxiliary | |
| CN109376736A (en) | A kind of small video target detection method based on depth convolutional neural networks | |
| CN107563299A (en) | A kind of pedestrian detection method using ReCNN integrating context informations | |
| CN111126155B (en) | Pedestrian re-identification method for generating countermeasure network based on semantic constraint | |
| CN118351487A (en) | Employee violation behavior identification method based on AI | |
| CN115410119A (en) | Violent movement detection method and system based on adaptive generation of training samples | |
| CN117079197B (en) | Intelligent building site management method and system | |
| KR102576747B1 (en) | System for local optimization of objects detector based on deep neural network and method for creating local database thereof | |
| CN113627383A (en) | Pedestrian loitering re-identification method for panoramic intelligent security | |
| CN117690086A (en) | Flood prevention gate identification and control method and system based on 5G and AI technology | |
| CN117475353A (en) | Video-based abnormal smoke identification method and system | |
| CN116152699B (en) | Real-time moving target detection method for hydropower plant video monitoring system | |
| CN115830701A (en) | Human violation behavior prediction method based on small sample learning | |
| CN116612355A (en) | Training method and device for face fake recognition model, face recognition method and device | |
| CN116740627A (en) | Violation warning method based on operation knowledge | |
| CN114387484B (en) | Improved mask wearing detection method and system based on yolov4 |
Legal Events
| 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 |