CN115223087A - Group control elevator traffic mode identification method - Google Patents
Group control elevator traffic mode identification method Download PDFInfo
- Publication number
- CN115223087A CN115223087A CN202210543435.5A CN202210543435A CN115223087A CN 115223087 A CN115223087 A CN 115223087A CN 202210543435 A CN202210543435 A CN 202210543435A CN 115223087 A CN115223087 A CN 115223087A
- Authority
- CN
- China
- Prior art keywords
- model
- elevator
- network
- data
- group control
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
- B66B5/0006—Monitoring devices or performance analysers
- B66B5/0012—Devices monitoring the users of the elevator system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for identifying a group control elevator traffic mode, relates to an elevator mode identification method, and relates to a method for constructing a passenger traffic condition detection model by adopting a yolov5 target detection technology, detecting passenger traffic through the model, extracting passenger traffic data and finishing the identification of the traffic mode and the scheduling optimization of elevator dispatching. The traffic pattern recognition in the invention can recognize real-time, effective and accurate pedestrian volume data, analyze the real pedestrian volume and distribute the traffic pattern according with the current pedestrian volume condition. The improved image processing technology is added into the traditional elevator group control system, and the traffic mode identification module is designed according to the actual and effective people flow condition, so that the people flow condition identified by the module has timeliness and authenticity, and the method has significance for the elevator group control system.
Description
Technical Field
The invention relates to an elevator mode identification method, in particular to a method for constructing a traffic mode model aiming at the operation of a group control elevator.
Background
Elevators have been developed over 100 years from the earliest mechanical control modes to motor traction control and now intelligent control, and elevator control has also gradually tended to mature, particularly after the advent of PLC. Modern building construction and planning are not single like housing or office, much more, the trend is towards commercialization and diversification, the floors are higher and higher, the business models are more and more, and the people flow conditions in the building are larger and larger, so that the elevator in the traditional sense is difficult to meet the requirements of the modern building.
Elevators are used as main transportation means inside buildings, and are responsible for transporting people up and down inside the buildings. Elevators appeared in the field of vision of people at the earliest in a mechanical manner, and single-step elevators based on electrical control became mainstream later with the development of electrical control. However, the single-step elevator is difficult to meet the requirement of passenger transportation when facing a building with large passenger flow and high floor number, so the group control elevator replaces the single-step elevator step by step, various performance indexes of the group control elevator are important for passengers, and the performance of a group control system can determine the waiting time and the boarding time of the passengers.
Elevators are non-linear, discrete and uncertain, so it is difficult to build an accurate mathematical model, and based on this, how to satisfy the dispatch of a group control elevator system will be a major problem to improve the performance of the group control system.
Disclosure of Invention
The invention aims to provide a method for identifying a group control elevator traffic mode, which takes target detection as an entry point, obtains the actual passenger traffic flow condition by detecting the elevator car inner cloth and elevator waiting personnel in real time, obtains the traffic mode according with the current traffic flow by integrating the traffic flow data by a control module in a system, and improves the operating efficiency of a group control system.
The purpose of the invention is realized by the following technical scheme:
a group control elevator traffic mode identification method comprises the steps of constructing an image detection network model for traffic mode identification, extracting data from the image model by a traffic mode identification module, and distributing traffic modes according to real-time conditions, and specifically comprises the following steps:
1) Making a passenger data set for elevator group control system target detection;
2) Adopting a deep learning pytorech frame to configure the environment of the network, and completing the model construction of yolov5 in the environment;
3) A CA attention mechanism is added into the original model frame of yolov5 to optimize the model;
4) Taking the preprocessed data set as the input of a network, training, loading yolov5s pre-training weight, and taking the CIOU as a loss function;
5) Detecting the network model by shooting videos of elevator passengers;
6) The traffic mode recognition module is used for sorting the pedestrian flow conditions detected by the system and deciding a proper elevator traffic mode according to the pedestrian flow data.
The group control elevator traffic mode identification method comprises the steps of preprocessing a picture data set of an input model; aiming at the identification of the elevator group control traffic mode, the data collection adopts the picture data of passengers when taking the elevator or waiting the elevator; collecting videos of monitoring equipment and performing framing processing on the videos to obtain images of passengers under various conditions, collecting and sorting elevator passenger pictures and converting the pictures into a JPG format, labeling picture data by adopting a labelimg tool, and outputting a label format in an xml format, wherein a label format of yolo is txt, so that a label in the xml format needs to be converted into the txt format through a code; when the network is trained, the data needs to be divided into a training set and a test set, wherein 80% of the data set is used as the training set of the network, and the rest 20% of the data set is used as the test set.
According to the method for identifying the group control elevator traffic mode, the picture data set is preprocessed, and the construction of an image processing target detection model is completed; the model architecture is divided into an input end, a backhaul Backbone network, a Neck network layer and a Head output end, wherein the input end processes an input picture, the speed of model training and the accuracy of the network are improved by adopting the modes of Mosaic data enhancement, self-adaptive anchor frame calculation, self-adaptive picture scaling and the like, the backhaul network adopts a Focus structure and a CSP structure to build a neural network of image characteristics, the Neck network adopts an FPN + PAN structure to strengthen the network characteristic fusion capability, and the Head adopts CIOU as a loss function to predict the image characteristics to generate a bounding box and a category confidence coefficient; the model loads yolov5s pre-training weight, the initial learning rate is 0.01, the momentum is set to 0.957, the loss gain is set to 0.53, the loss function is CIOU, and the other parameters are default values.
According to the method for identifying the group control elevator traffic mode, the target detection model is built, in order to further improve the accuracy of the model, a CA attention mechanism is adopted to optimize the network model, an attention mechanism is added in the built yolov5 trunk network, the mechanism embeds position information into channel attention to obtain the attention weight of a feature map in height and width, and finally the network detection accuracy is enhanced by integrating the attention weight into the original feature map; the CA attention mechanism formula is as follows:
compared with an initial model, the model added with the CA attention mechanism has the advantage that map is increased by about 0.5%, and the model has significance for detecting passenger traffic flow.
The method for identifying the group control elevator traffic mode comprises the steps of completing optimization of a model, setting parameters such as the number of data input into a network at one time, the number of training rounds, a working thread and the like, and starting to train the model; after the model training is finished, checking whether the performance index of the model is reasonable, starting a camera to detect input data, and checking the confidence coefficient of a prediction box.
The model is trained and tested, three ten-floor elevator group control systems use a target detection network to extract data of passengers, the system can record the number of elevator waiting persons of each floor and the number of elevator taking persons in an elevator car, record the data of the average elevator taking time, the average elevator waiting time, the staying time of a target floor and the like of the passengers in a certain interval time, and transmit the data to an information management module for optimizing elevator dispatching; the traffic mode identification module can obtain the pedestrian flow data detected by the target, and selects a traffic mode suitable for the current situation according to the pedestrian flow data detected in real time.
The invention has the advantages and effects that:
1. the invention provides an application of a yolov 5-based target detection algorithm in group control elevator traffic pattern recognition. The yolov5 target detection algorithm is adopted to detect passengers in the elevator car and passengers waiting for the elevator, real-time passenger flow conditions are obtained, data are transmitted to the corresponding control modules, and finally the system can select the traffic mode suitable for the current traffic mode according to the passenger flow data. Compared with the traditional time series model, the target detection technology has high efficiency, accuracy and real-time performance, the system can obtain current actual effective people flow data, the accuracy of traffic pattern recognition can be improved, and the elevator dispatching strategy is improved finally.
2. The invention adopts a method based on target detection, adds image identification into an elevator group control system, can obtain the passenger flow conditions in real time by detecting passengers inside and outside a real car, has extremely strong authenticity and timeliness compared with the traditional time series prediction algorithm, and can decide the traffic mode which best accords with the current passenger flow by transmitting the actual passenger flow data to the traffic mode identification module. In addition, the average elevator taking time and the average elevator waiting time of the passengers can be detected to optimize the dispatching algorithm. A CA attention mechanism is added into the yolov5 target detection algorithm, the original algorithm is optimized, the network precision is improved, and the traffic mode identification of the system is facilitated.
Drawings
FIG. 1 is a functional structure diagram of a group control system for three ten-storey elevators according to the present invention;
FIG. 2 is a view of the yolov5 model architecture of the present invention;
FIG. 3 is a diagram of a CA attention mechanism network of the present invention;
FIG. 4 is a flow chart of the model training of the present invention;
FIG. 5 is a map accuracy map of the present invention;
FIG. 6 is a graph of the loss function of the present invention;
fig. 7 is a flow chart of the present invention.
Detailed Description
The invention relates to an application of a yolov 5-based target detection algorithm in group control elevator traffic pattern recognition, which comprises the following steps:
1. making a passenger data set for elevator group control system target detection;
2. adopting a deep learning pytorech frame to configure the environment of the network, and completing the model construction of yolov5 in the environment;
3. a CA attention mechanism is added into the original model frame of yolov5 to optimize the model;
4. taking the preprocessed data set as the input of a network, training, loading yolov5s pre-training weight, and taking the CIOU as a loss function;
5. detecting the network model by shooting videos of elevator passengers;
6. and the traffic mode identification module is used for sorting the pedestrian flow conditions detected by the system and deciding a proper elevator traffic mode according to the pedestrian flow data.
The specific implementation steps of the step 1 are as follows:
collecting pictures of elevator passengers, and labeling the pictures by using labelimg;
converting the marked data into a txt format of yolo;
the data set is divided into a training set and a test set.
The specific implementation steps of the step 2 are as follows:
importing a configuration environment required by a project;
and constructing a yolo model, wherein the main bodies of the yolo model comprise an Input end, a Backbone network of Backbone, a Neck network layer and a Head output end.
The concrete implementation steps of the step 3 are as follows:
a CA (Coordinate Attention) Coordinate Attention mechanism is introduced to strengthen the optimization of the network and improve the accuracy of the network; the CA comprises two parts, coordinate information is embedded, coordinate attention is generated, the CA module divides an input feature graph into two directions of width and height to perform global average pooling respectively, feature output in the two directions is obtained, and an output formula is as follows:
（1）
wherein c is a channel, h is a height, w is a width, and X is an input characteristic diagram; splicing the feature maps in the two directions, and converting the feature maps into 1 × 1 shared convolution after completing to obtain a new feature output f, wherein the formula of the f is as follows:
wherein, the first and the second end of the pipe are connected with each other,
in order to activate the function(s),
the feature map is subjected to normalization processing; after f is obtained, it needs to be sliced into two separate tensors along the spatial dimension
And
using convolution of two 1 x 1 s
And
and (3) converting the two tensors into the same channel number as the input feature map X, and finally obtaining the attention weights of the feature map on the width and the height respectively through a sigmoid activation function, wherein the attention weight formula is as follows:
after the attention weights in the width direction and the height direction are obtained, weighting calculation is carried out on the original feature map, finally, the feature map with the attention weights in the width direction and the height direction is obtained, and finally, the output formula is as follows:
the feature map obtains attention weights in the width direction and the height direction, and the accuracy of the model is improved.
The specific implementation steps of the step 4 are as follows:
loading a pre-training weight of yolov5s, and improving the speed and precision of network training;
the preprocessed data set is divided into a training set and a test set and sent to a network;
setting the bath size to be 16, the epochs to be 100, the input size to be 640, and selecting default setting for the hyperparameters;
the Precision and the Recall rate Recall are selected as indexes for measuring the model, and the formulas are respectively as follows:
wherein TP is that positive samples are correctly identified, FP is that negative samples are identified as positive samples, and FN is that negative samples are correctly identified; in order to solve the divergence problem of the IOU in the training process, the CIOU is used as a loss function of the Bounding box, the CIOU considers the distance, the overlapping rate, the scale and the penalty factors between the target and the anchor, the regression stability of the target frame is enhanced, and the regression convergence speed of the prediction frame and the real frame is accelerated, and the formula is as follows:
wherein, the first and the second end of the pipe are connected with each other,
in order to be a function of the weight,
to predict the aspect ratio of the box to the real box,
c is the Euclidean distance between the central points of the prediction frame and the real frame, and the diagonal distance which can simultaneously contain the minimum closure area of the prediction frame and the real frame; finally substitute into
A loss function is obtained, which is formulated as follows:
the concrete implementation steps of the step 5 are as follows:
detecting performance indexes of Precision, recall, F1 score, map, loss function and the like of the training model; and inputting the passenger video in the real scene into the network, and detecting the confidence coefficient of network identification.
The concrete implementation steps of the step 6 are as follows:
the system extracts passenger data in front of each layer of car door and passenger data in the car to obtain real-time specific passenger flow rate conditions, and calculates the waiting time and the taking time of passengers and the time of the elevator staying at a certain floor under different flow rates; the data are transmitted to a data management module, and the data management module arranges the data and respectively transmits the data to corresponding data modules; the traffic mode identification module can obtain the pedestrian flow condition of each floor, and selects a proper traffic mode according to the proportion of the current passengers; when the passengers on the lower floor are more and the uplink call signal gives a priority response, the system can select an uplink peak traffic mode; when more passengers exist on the high floor and the downlink calling signal gives priority to the response, the system can select a downlink peak traffic mode; when the passengers are scattered on all floors, the system can select an interlayer traffic mode; when the number of passengers is small, the system selects the idle traffic mode.
Examples
The invention uses a target detection method to be applied to the traffic mode recognition of an elevator group control system, takes three ten-layer elevator group control systems as an example, the system is divided into two parts, namely a target detection algorithm network and the traffic mode recognition, and the structure diagram of the three ten-layer elevator group control system is shown in figure 1.
A pre-processing operation of the input picture data set is required before the network is built. Firstly, the pictures of the elevator passengers are collected and sorted, the pictures are converted into a JPG format, then picture data are marked by a labellimg tool, then a label format is output in an xml format, and a label in the xml format needs to be converted into a txt format through codes because the label format of yolo adopts txt. When the network is trained, the data needs to be divided into a training set and a test set, wherein 80% of the data set is used as the training set of the network, and the rest 20% of the data set is used as the test set.
After the preprocessing of the data set is completed, a network model of yolov5 algorithm needs to be built and the running environment needs to be configured. The method includes the steps that pycharm is used as an IDE (integrated development environment) of a project, a downloaded pytorch environment is guided into the project by using conda, parameter setting and model building are needed for yolov5 after environment configuration is completed, and a yolov5 model is mainly provided with an Input end, a Backbone network of a Backbone, a neutral network layer and a Head output end as shown in FIG. 2. The input end processes the input picture, and the model training speed and the network precision are improved by adopting the modes of Mosaic data enhancement, self-adaptive anchor frame calculation, self-adaptive picture scaling and the like; the Backbone network adopts a Focus structure and a CSP structure to build a neural network of image characteristics; the Neck network adopts an FPN + PAN structure to strengthen the network feature fusion capability; and the Head predicts the image characteristics by adopting the CIOU as a loss function to generate a bounding box and a category confidence coefficient.
After yolov5 initial model building is completed, a CA attention mechanism is added into a backbone network of the yolov5 initial model building to optimize the network model. CA is a novel mobile network attention mechanism, and can improve the accuracy of the network. The CA attention mechanism module diagram is shown in FIG. 3, the CA is added into a yolov5 model, the original model framework is modified, and in the first step, a CA function is written into common to complete the model structure of the CA; second step, modification of CA model is performed in yolo; and thirdly, rewriting a Backbone network of the yolov5s model, and adding the CA into the Backbone network model.
After the model is improved, the preprocessed picture data set needs to be input into a network for training. The pre-training weight of yolov5s is added before the model is trained to optimize the training of the model, shorten the time of network training and improve the accuracy of model prediction. Setting the path, the type and the number of a training data set when parameters are changed for a training model; loading a pre-training weight type and a path of the model; selecting hyper-parameters and activation functions which accord with the training types of the data set; setting the training working thread, the number of pictures input into the network and the number of training rounds. After the model parameters are set, the model starts to be trained to obtain the model weight according with the system, and the training process of the model is shown in fig. 4.
And (3) after the model training is finished, checking whether the model training is good or bad by adopting performance indexes such as Precision, recall, F1 score, map and the like, wherein the map Precision of the model is shown in figure 5, and the loss function is shown in figure 6. The video of the elevator passenger is used as the input of the model for detecting the accuracy of the prediction frame of the model recognition passenger.
The system can record the number of elevator waiting persons on each floor and the number of elevator taking persons in the elevator car, record the average elevator taking time, the average elevator waiting time, the residence time of the target floor and other data of the passengers in a certain interval time, and transmit the data to the information management module for optimizing the elevator dispatching. The traffic mode identification module can obtain the pedestrian flow data detected by the target, and selects a traffic mode suitable for the current situation according to the pedestrian flow data detected in real time. The identification of the elevator traffic mode is realized by a target detection method, and real people flow data can be effectively detected in real time.
Claims (6)
1. A group control elevator traffic mode identification method is characterized by comprising the following steps of constructing an image detection network model for traffic mode identification, extracting data from the image model by a traffic mode identification module, and distributing traffic modes according to real-time conditions:
1) Making a passenger data set for elevator group control system target detection;
2) Adopting a deep learning pytorech frame to configure the environment of the network, and completing the model construction of yolov5 in the environment;
3) A CA attention mechanism is added into the original model frame of yolov5 to optimize the model;
4) Taking the preprocessed data set as the input of a network, training, loading yolov5s pre-training weight, and taking the CIOU as a loss function;
5) Detecting the network model by shooting videos of elevator passengers;
6) The traffic mode recognition module is used for sorting the pedestrian flow conditions detected by the system and deciding a proper elevator traffic mode according to the pedestrian flow data.
2. A method for group control elevator traffic pattern recognition according to claim 1, wherein the system includes preprocessing the picture data set of the input model; aiming at the identification of the elevator group control traffic mode, the data collection adopts the picture data of passengers when taking the elevator or waiting the elevator; collecting videos of monitoring equipment and performing framing processing on the videos to obtain images of passengers under various conditions, collecting and sorting elevator passenger pictures, converting the pictures into a JPG format, labeling picture data by adopting a labelimg tool, outputting a label format in an xml format, and converting a label in an xml format into a txt format by adopting a code because the label format of yolo is txt; when the network is trained, the data needs to be divided into a training set and a test set, wherein 80% of the data set is used as the training set of the network, and the rest 20% of the data set is used as the test set.
3. The method for group control elevator traffic pattern recognition according to claim 2, wherein the picture data set is preprocessed to complete construction of an image processing target detection model; the model architecture is divided into an input end, a Backbone network of a backhaul, a neutral network layer and a Head output end, wherein the input end processes an input picture, the speed of model training and the accuracy of the network are improved in the forms of Mosaic data enhancement, self-adaptive anchor frame calculation, self-adaptive picture scaling and the like, the Backbone network adopts a Focus structure and a CSP structure to build a neural network of image characteristics, the neutral network adopts an FPN + PAN structure to enhance the network characteristic fusion capability, and the Head adopts CIOU as a loss function to predict the image characteristics to generate a boundary frame and a category confidence; the model loads yolov5s pre-training weight, the initial learning rate is 0.01, the momentum is set to be 0.957, the loss gain is set to be 0.53, the loss function is CIOU, and the other parameters are default values.
4. The method for identifying the traffic mode of the group-controlled elevator according to claim 3, wherein the target detection model is built, in order to further improve the accuracy of the model, a network model is optimized by adopting a CA attention mechanism, an attention mechanism is added in the built yolov5 backbone network, the mechanism embeds position information into channel attention to obtain the attention weight of a feature map on the height and the width, and finally the network detection accuracy is enhanced by integrating the attention weight into an original feature map; the CA attention mechanism is as follows:
（1）
compared with an initial model, the model added with the CA attention mechanism has the advantage that map is increased by about 0.5%, and the model has significance for detecting passenger flow.
5. The method for group control elevator traffic pattern recognition according to claim 4, wherein the optimization of the model is completed, parameters such as the number of data sheets input into the network at one time, the number of training rounds, the working thread and the like are set, and the model is trained; after the model training is finished, whether the performance index of the model is reasonable or not is checked, the camera is started to detect input data, and the confidence coefficient of the prediction box is checked.
6. The method for group control elevator traffic pattern recognition according to claim 5, wherein the model training is completed and tested, the data of passengers are extracted by the three ten-floor elevator group control system by using a target detection network, the system records the number of elevator waiting persons on each floor and the number of elevator passengers in the elevator car, records the data of the average elevator waiting time, the staying time of the target floor and the like of the passengers in a certain interval time, and transmits the data to the information management module for optimizing elevator scheduling; the traffic pattern recognition module can obtain the pedestrian flow data detected by the target, and the traffic pattern recognition module selects the current pedestrian flow data according to the real-time detection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210543435.5A CN115223087A (en) | 2022-05-19 | 2022-05-19 | Group control elevator traffic mode identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210543435.5A CN115223087A (en) | 2022-05-19 | 2022-05-19 | Group control elevator traffic mode identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115223087A true CN115223087A (en) | 2022-10-21 |
Family
ID=83608119
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210543435.5A Pending CN115223087A (en) | 2022-05-19 | 2022-05-19 | Group control elevator traffic mode identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115223087A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115546734A (en) * | 2022-11-25 | 2022-12-30 | 常熟理工学院 | Elevator people flow visual statistical method and system based on deep learning and storage medium |
-
2022
- 2022-05-19 CN CN202210543435.5A patent/CN115223087A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115546734A (en) * | 2022-11-25 | 2022-12-30 | 常熟理工学院 | Elevator people flow visual statistical method and system based on deep learning and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110262463B (en) | Rail transit platform door fault diagnosis system based on deep learning | |
| CN112069868A (en) | Unmanned aerial vehicle real-time vehicle detection method based on convolutional neural network | |
| CN100595121C (en) | Target place optimized dispatching method and system | |
| CN111784685A (en) | Power transmission line defect image identification method based on cloud edge cooperative detection | |
| CN111079640B (en) | Vehicle type identification method and system based on automatic amplification sample | |
| CN109766769A (en) | A kind of road target detection recognition method based on monocular vision and deep learning | |
| CN108764456B (en) | Airborne target identification model construction platform, airborne target identification method and equipment | |
| WO2023216721A1 (en) | Concrete dam defect time sequence image intelligent identification method | |
| CN112287827A (en) | Complex environment pedestrian mask wearing detection method and system based on intelligent lamp pole | |
| CN109935080A (en) | The monitoring system and method that a kind of vehicle flowrate on traffic route calculates in real time | |
| CN112347864A (en) | Method, device, equipment and system for sensing and inducing rail transit passenger flow | |
| CN115719475B (en) | Three-stage trackside equipment fault automatic detection method based on deep learning | |
| CN111540203B (en) | Method for adjusting green light passing time based on fast-RCNN | |
| CN115223087A (en) | Group control elevator traffic mode identification method | |
| CN103605960B (en) | A kind of method for identifying traffic status merged based on different focal video image | |
| CN117237884A (en) | Interactive inspection robot based on berth positioning | |
| CN116824482A (en) | High-precision elevator traffic pattern recognition method for target detection | |
| CN116229424A (en) | Intelligent networking automobile pedestrian detection method based on improved YOLO v3 | |
| CN115937788A (en) | Yolov5 industrial area-based safety helmet wearing detection method | |
| CN111339926A (en) | Intelligent recognition system for behavior specification of personnel in electric power business hall | |
| CN113963310A (en) | People flow detection method and device for bus station and electronic equipment | |
| CN112116241B (en) | Intelligent public traffic scheduling method | |
| CN113421067A (en) | Shared bicycle brand identification and quantity management method and system | |
| CN113158835A (en) | Traffic accident intelligent detection method based on deep learning | |
| CN113706860A (en) | Intelligent timing traffic light control method based on raspberry group |
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 |