CN113378805B - Height limiting device detection method and system based on deep learning and intelligent terminal - Google Patents
Height limiting device detection method and system based on deep learning and intelligent terminal Download PDFInfo
- Publication number
- CN113378805B CN113378805B CN202110927771.5A CN202110927771A CN113378805B CN 113378805 B CN113378805 B CN 113378805B CN 202110927771 A CN202110927771 A CN 202110927771A CN 113378805 B CN113378805 B CN 113378805B
- Authority
- CN
- China
- Prior art keywords
- limiting device
- height limiting
- height
- detection
- detection model
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a method and a system for detecting a height limiting device based on deep learning and an intelligent terminal, wherein the method comprises the following steps: constructing a data set of the height limiting device; training the characteristics of the height limiting device in the data set by using a deep learning network to obtain a detection model of the height limiting device; testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model; and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information. The technical problem that the position of a height limiting device in the prior art cannot be accurately acquired is solved.
Description
Technical Field
The invention relates to the technical field of auxiliary driving, in particular to a height limiting device detection method and system based on deep learning and an intelligent terminal.
Background
With the continuous development of automobile technology, the number of vehicles with higher vehicle body heights, such as large buses, large trucks, motor homes and the like, continuously increases. However, accidents such as casualties and property loss due to collisions between vehicles and height-limiting devices occur frequently due to negligence of drivers of large vehicles or failure to accurately estimate the positions of the height-limiting devices. Therefore, a method and a system for detecting a height limiting device based on deep learning and an intelligent terminal are provided to improve the position detection accuracy of the height limiting device, avoid the occurrence of collision accidents and guarantee the driving safety, so that the problem to be solved by technical personnel in the field is solved urgently.
Disclosure of Invention
Therefore, the embodiment of the invention provides a method and a system for detecting a height limiting device based on deep learning and an intelligent terminal, so as to solve the technical problem that the position of the height limiting device in the prior art cannot be accurately acquired.
In order to achieve the above object, the embodiments of the present invention provide the following technical solutions:
a method for detecting a height limiting device based on deep learning, the method comprising:
constructing a data set of the height limiting device;
training the characteristics of the height limiting device in the data set by using a deep learning network to obtain a detection model of the height limiting device;
testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model;
and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information.
Further, the constructing the data set of the height limiting device specifically includes:
using a calibrated binocular camera to collect all height limiting devices encountered in a preset road section in the real driving environment of the automobile;
in the acquisition process, the automobile is ensured to run forwards at a preset speed, and all image frames from the imaging of the height limiting device in the binocular camera to the time when the automobile cannot acquire the height limiting data through the height limiting device are acquired;
all image frames are formed into a data set.
Further, the data set is divided into a training set, a verification set and a test set according to a preset proportion.
Further, the constructing the data set of the height limiting device further comprises:
marking all height limiting devices in the collected image by using a rectangular frame; wherein the content of the first and second substances,
the marked rectangular frame completely wraps the height limiting part of the height limiting device and does not contain objects except the height limiting device;
marking all height limiting devices contained in the image, and marking a plurality of height limiting devices through rectangular frames in a one-to-one correspondence manner;
a label is embedded for each labeled height limiting device, the label including a category and a name of the height limiting device.
Further, the training for the features of the height limiting device in the data set by using the deep learning network to obtain a detection model of the height limiting device specifically includes:
the deep learning network is a YOLOV5 network;
inputting the training set of the collected height-limiting data and the labeled corresponding label into a YOLOV5 network for training;
and observing the loss of the training process, and stopping training when the loss of the model is stable so as to obtain the height limiting device detection model.
Further, the testing the detection model of the height limiting device specifically includes:
inputting the collected test set of the height limiting data into the detection model of the height limiting device for testing;
acquiring position information of an enclosing frame output by a detection model of the height limiting device, and visualizing the information of the enclosing frame to obtain the position of the height limiting device;
acquiring the detection confidence of the height limiting device, and judging the reliability of detection according to the numerical value of the detection confidence;
if the detection confidence is judged to be greater than or equal to the confidence threshold, taking the detection result corresponding to the detection confidence as the actual detection result;
and if the detection confidence is judged to be smaller than the confidence threshold, the model is retrained.
Further, the performing error analysis on the detection model of the height limiting device specifically includes:
carrying out error analysis according to the input height limiting device test set data;
the average precision mean value and the transmission frame number per second are selected as evaluation indexes of a detection algorithm of the height limiting device, and the calculation formula of the precision is as follows:
accuracy = number of correct positive samples detected by current traversal/number of all positive samples detected by current traversal;
the recall ratio is calculated by the formula:
recall = number of correct positive samples/number of all true values detected by the current traversal.
The invention also provides a height limiting device detection system based on deep learning, which comprises:
the data set construction unit is used for constructing a data set of the height limiting device;
the model training unit is used for training the characteristics of the height limiting device in the data set by utilizing a deep learning network to obtain a detection model of the height limiting device;
the model testing unit is used for testing and analyzing errors of the height limiting device detection model to obtain an adjusted height limiting device detection model;
and the result output unit is used for acquiring the height limiting device on the current driving road based on the adjusted height limiting device detection model and outputting early warning information.
The present invention also provides an intelligent terminal, including: the device comprises a data acquisition device, a processor and a memory;
the data acquisition device is used for acquiring data; the memory is to store one or more program instructions; the processor is configured to execute one or more program instructions to perform the method as described above.
The invention provides a height limiting device detection method based on deep learning, which comprises the steps of constructing a data set of a height limiting device, and training characteristics of the height limiting device in the data set by utilizing a deep learning network to obtain a detection model of the height limiting device; testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model; and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information. The method is based on a deep learning method to detect a height limiting device, and finally carries out error analysis to judge the quality of a model through data acquisition and marking, model training and testing and evaluation indexes. The method can effectively detect the specific position information of the height limiting device and provide data support for judging whether the height limiting device can pass through. Therefore, the position detection accuracy of the height limiting device is improved, the occurrence of collision accidents is avoided, the driving safety is guaranteed, and the technical problem that the position of the height limiting device cannot be accurately acquired in the prior art is solved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
The structures, ratios, sizes, and the like shown in the present specification are only used for matching with the contents disclosed in the specification, so as to be understood and read by those skilled in the art, and are not used to limit the conditions that the present invention can be implemented, so that the present invention has no technical significance, and any structural modifications, changes in the ratio relationship, or adjustments of the sizes, without affecting the effects and the achievable by the present invention, should still fall within the range that the technical contents disclosed in the present invention can cover.
FIG. 1 is a flowchart of an embodiment of a method for detecting a height limiter based on deep learning according to the present invention;
fig. 2 is a schematic diagram of a network structure of YOLOV 5;
FIG. 3 is a diagram showing the results of bridge opening detection;
FIG. 4 is a diagram showing the result of the detection of the height-limiting rod;
fig. 5 is a block diagram of an embodiment of a deep learning-based height limiter detection system according to the present invention.
Detailed Description
The present invention is described in terms of particular embodiments, other advantages and features of the invention will become apparent to those skilled in the art from the following disclosure, and it is to be understood that the described embodiments are merely exemplary of the invention and that it is not intended to limit the invention to the particular embodiments disclosed. 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.
In one embodiment, as shown in fig. 1, the method for detecting a height limiting device based on deep learning provided by the present invention includes the following steps:
s1: constructing a data set of the height limiting device; specifically, a calibrated binocular camera is used for collecting all height limiting devices encountered in a preset road section in the real driving environment of the automobile; in the acquisition process, the automobile is ensured to run forwards at a preset speed, and all image frames from the imaging of the height limiting device in the binocular camera to the time when the automobile cannot acquire the height limiting data through the height limiting device are acquired; all image frames are formed into a data set. In order to facilitate subsequent training and testing, the data set is divided into a training set, a verification set and a testing set according to a preset proportion.
In a specific use scene, the calibrated binocular camera is used, the height limiting devices (such as a height limiting rod, a bridge opening, a memorial archway, an unconventional height limiting device and the like) encountered in the real driving environment of the automobile are collected, the automobile is guaranteed to run forwards at a normal speed in the collection process, and the height limiting data cannot be collected by the automobile passing through the height limiting devices from the beginning of imaging of the height limiting devices on the binocular camera. The collected images are divided into a training set, a verification set and a test set according to the ratio of 6:2: 2.
S2: and training the characteristics of the height limiting device in the data set by using a deep learning network to obtain a detection model of the height limiting device.
There are two hard requirements for the choice of deep learning networks for the height-limited device detection problem. The method comprises the steps of firstly requiring the forward reasoning time of a deep learning network to meet the requirement of real-time performance so as to meet the landability of a detection algorithm of a height limiting device. Secondly, the model trained by the deep learning network is required to be efficient enough, the occurrence of false detection and missed detection is reduced as far as possible, and therefore the practicability of the detection algorithm of the height limiting device is improved. Based on the above requirements, the YOLOV5 network is selected as the deep learning network of the height-limiting device. A schematic diagram of a network structure of YOLOV5 is shown in fig. 2. In fig. 2, CONV represents a convolution module, the CBL module is formed by combining a convolution layer, batch normalization, and a LeakyRelu function, the Focus module is formed by combining a module formed by slicing and splicing 4 slice modules and the CBL module, the CSP module is formed by combining the CBL module, the convolution layer, the batch normalization, and the LeakyRelu function, and the SPP module is formed by combining the CBL module, a maximum pooling layer, and the CBL module.
Based on this, the training is performed for the features of the height limiting device in the data set by using the deep learning network to obtain a detection model of the height limiting device, which specifically includes:
the deep learning network is a YOLOV5 network;
inputting the training set of the collected height-limiting data and the labeled corresponding label into a YOLOV5 network for training;
and calculating the inference result and the labeled result of the network through a loss function in the training process to obtain a loss value, observing the loss value in the training process, and stopping training when the loss value of the model is stable to obtain the height limiting device detection model. Specifically, the prediction result and the labeling result are input into a loss function for calculation in the training process, so as to obtain a specific loss value. The loss value will go through a high-to-low transition, and is said to be stable when the loss value no longer decreases.
S3: and testing and analyzing errors of the height limiting device detection model to obtain an adjusted height limiting device detection model.
After model training is completed, testing the height limiting device detection model obtained by training, and specifically comprising the following steps:
s31: inputting the collected test set of the height limiting data into the detection model of the height limiting device for testing;
s32: acquiring position information of an enclosing frame output by a detection model of the height limiting device, and visualizing the information of the enclosing frame to obtain the position of the height limiting device;
s33: acquiring the detection confidence of the height limiting device, and judging the reliability of detection according to the numerical value of the detection confidence;
s34: if the detection confidence is judged to be greater than or equal to the confidence threshold, taking the detection result corresponding to the detection confidence as the actual detection result;
s35: and if the detection confidence is judged to be smaller than the confidence threshold, the model is retrained.
And inputting the collected test set of the height limiting data into a stored height limiting device detection model for testing, outputting the position information of the surrounding frame by the model, and outputting the result of the visualized information of the surrounding frame in the form of an image. For example, as shown in fig. 3, the detection result of the bridge opening is shown, and as shown in fig. 4, the detection result of the height-limiting rod is shown, and the number after classification is the confidence of the detection, and the reliability of the detection can be judged according to the value of the confidence.
In order to improve the accuracy of the model, after the detection model of the height limiting device is tested, the detection model of the height limiting device is subjected to error analysis, and the method specifically comprises the following steps:
carrying out error analysis according to the input height limiting device test set data;
the average precision mean value and the transmission frame number per second are selected as evaluation indexes of a detection algorithm of the height limiting device, and the calculation formula of the precision is as follows:
accuracy = number of correct positive samples detected by current traversal/number of all positive samples detected by current traversal;
the recall ratio is calculated by the formula:
recall = number of correct positive samples/number of all true values detected by the current traversal.
In principle, the Average Precision Average (mean Average Precision) and the number of Frames Per Second (FPS) are selected as evaluation indexes of the height-limiting device detection algorithm. In target detection, a model usually detects many objects, and each class can draw an RP curve. Wherein the RP curve is a curve with recall rate on the abscissa and accuracy on the ordinate. The calculation method of the AP value of each category is to smooth the curve, and the maximum accuracy value on the right side of each point is taken and connected into a straight line. (also known as the interpolated AP method) then takes 11 points (one point per 0.1 on the recall axis) and directly averages the sum of the accuracies of the 11 points. And the average AP value of a plurality of categories is mAP, and the larger the mAP value is, the better the detection effect is.
S4: and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information.
Further, in order to facilitate subsequent model training, after constructing the data set of the height limiting device, the method further includes:
marking all height limiting devices in the collected image by using a rectangular frame; wherein the content of the first and second substances,
the marked rectangular frame completely wraps the height limiting part of the height limiting device and does not contain objects except the height limiting device;
marking all height limiting devices contained in the image, and marking a plurality of height limiting devices through rectangular frames in a one-to-one correspondence manner;
a label is embedded for each labeled height limiting device, the label including a category and a name of the height limiting device.
The acquired image needs to be labeled. The specific labeling is shown in fig. 3. Wherein, the marking process should satisfy the following three points:
the marked rectangular frame is required to completely wrap the height limiting part of the height limiting device and does not contain objects except the height limiting device as far as possible.
All height-limiting devices contained in the image should be labeled and labeled with a plurality of rectangular frames.
Each marked height limiting device should be marked with its category, and different kinds of height limiting rods should be named differently, such as height limiting rods, memorial archways, bridge openings and the like.
In a specific embodiment, the method for detecting the height limiting device based on deep learning provided by the invention comprises the steps of constructing a data set of the height limiting device, and training characteristics of the height limiting device in the data set by using a deep learning network to obtain a detection model of the height limiting device; testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model; and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information. The method is based on a deep learning method to detect a height limiting device, and finally carries out error analysis to judge the quality of a model through data acquisition and marking, model training and testing and evaluation indexes. The method can effectively detect the specific position information of the height limiting device and provide data support for judging whether the height limiting device can pass through. Therefore, the position detection accuracy of the height limiting device is improved, the occurrence of collision accidents is avoided, the driving safety is guaranteed, and the technical problem that the position of the height limiting device cannot be accurately acquired in the prior art is solved.
That is to say, the height limiting device is detected based on a deep learning method, the model is trained and tested through data acquisition and labeling, and finally the model is judged to be good or bad through error analysis of evaluation indexes. The method can effectively detect the specific position information of the height limiting device, so that the large automobile can judge that the large automobile can pass through the height limiting device or predict that the large automobile cannot pass through the height limiting device in advance according to the information, and further data support is provided for making further decisions.
In addition to the above method, the present invention further provides a system for detecting a height limiting device based on deep learning, and in one embodiment, as shown in fig. 5, the system includes:
a data set construction unit 100 for constructing a data set of the height-limiting device.
The data set construction unit 100 is specifically configured to collect, in the real driving environment of the automobile, all height limiting devices encountered in a preset road segment by using the calibrated binocular camera; in the acquisition process, the automobile is ensured to run forwards at a preset speed, and all image frames from the imaging of the height limiting device in the binocular camera to the time when the automobile cannot acquire the height limiting data through the height limiting device are acquired; all image frames are formed into a data set. And, the data set constructing unit 100 is further configured to divide the data set into a training set, a verification set and a test set according to a preset ratio.
And a model training unit 200, configured to train features of the height limiting device in the data set by using a deep learning network to obtain a height limiting device detection model. The deep learning network is a YOLOV5 network, and the model training unit 200 is specifically configured to input a training set of the acquired height-limiting data and a labeled corresponding label into the YOLOV5 network for training; and observing the loss of the training process, and stopping training when the loss of the model is stable so as to obtain the height limiting device detection model.
And the model testing unit 300 is configured to perform testing and error analysis on the height limiting device detection model to obtain an adjusted height limiting device detection model.
The model test unit 300 is specifically configured to:
inputting the collected test set of the height limiting data into the detection model of the height limiting device for testing;
acquiring position information of an enclosing frame output by a detection model of the height limiting device, and visualizing the information of the enclosing frame to obtain the position of the height limiting device;
acquiring the detection confidence of the height limiting device, and judging the reliability of detection according to the numerical value of the detection confidence;
if the detection confidence is judged to be greater than or equal to the confidence threshold, taking the detection result corresponding to the detection confidence as the actual detection result;
and if the detection confidence is judged to be smaller than the confidence threshold, the model is retrained.
And a result output unit 400, configured to obtain a height limiting device on the current driving road based on the adjusted height limiting device detection model, and output early warning information.
Further, the system further includes a feature labeling unit 500, where the feature labeling unit 500 is specifically configured to: marking all height limiting devices in the collected image by using a rectangular frame; wherein the content of the first and second substances,
the marked rectangular frame completely wraps the height limiting part of the height limiting device and does not contain objects except the height limiting device;
marking all height limiting devices contained in the image, and marking a plurality of height limiting devices through rectangular frames in a one-to-one correspondence manner;
a label is embedded for each labeled height limiting device, the label including a category and a name of the height limiting device.
In the above specific embodiment, the height limiting device detection system based on deep learning provided by the invention is configured to build a data set of a height limiting device, and train the characteristics of the height limiting device in the data set by using a deep learning network to obtain a height limiting device detection model; testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model; and acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information. The method is based on a deep learning method to detect a height limiting device, and finally carries out error analysis to judge the quality of a model through data acquisition and marking, model training and testing and evaluation indexes. The method can effectively detect the specific position information of the height limiting device and provide data support for judging whether the height limiting device can pass through. Therefore, the position detection accuracy of the height limiting device is improved, the occurrence of collision accidents is avoided, the driving safety is guaranteed, and the technical problem that the position of the height limiting device cannot be accurately acquired in the prior art is solved.
The present invention also provides an intelligent terminal, including: the device comprises a data acquisition device, a processor and a memory;
the data acquisition device is used for acquiring data; the memory is to store one or more program instructions; the processor is configured to execute one or more program instructions to perform the method as described above.
In correspondence with the above embodiments, embodiments of the present invention also provide a computer storage medium containing one or more program instructions therein. Wherein the one or more program instructions are for executing the method as described above by a binocular camera depth calibration system.
In an embodiment of the invention, the processor may be an integrated circuit chip having signal processing capability. The Processor may be a general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete Gate or transistor logic device, discrete hardware component.
The various methods, steps and logic blocks disclosed in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The processor reads the information in the storage medium and completes the steps of the method in combination with the hardware.
The storage medium may be a memory, for example, which may be volatile memory or nonvolatile memory, or which may include both volatile and nonvolatile memory.
The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable PROM (EEPROM), or a flash Memory.
The volatile Memory may be a Random Access Memory (RAM) which serves as an external cache. By way of example and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), SLDRAM (SLDRAM), and Direct Rambus RAM (DRRAM).
The storage media described in connection with the embodiments of the invention are intended to comprise, without being limited to, these and any other suitable types of memory.
Those skilled in the art will appreciate that the functionality described in the present invention may be implemented in a combination of hardware and software in one or more of the examples described above. When software is applied, the corresponding functionality may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.
The above embodiments are only for illustrating the embodiments of the present invention and are not to be construed as limiting the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made on the basis of the embodiments of the present invention shall be included in the scope of the present invention.
Claims (8)
1. A method for detecting a height limiting device based on deep learning is characterized by comprising the following steps:
constructing a data set of a height limiting device, wherein the height limiting device comprises a height limiting rod, a bridge opening and a memorial archway;
training the characteristics of the height limiting device in the data set by using a deep learning network to obtain a detection model of the height limiting device;
testing and error analyzing the height limiting device detection model to obtain an adjusted height limiting device detection model;
acquiring a height limiting device on the current driving road based on the adjusted height limiting device detection model, and outputting early warning information;
the testing of the height limiting device detection model specifically includes:
inputting the collected test set of the height limiting data into the detection model of the height limiting device for testing;
acquiring position information of an enclosing frame output by a detection model of the height limiting device, and visualizing the information of the enclosing frame to obtain the position of the height limiting device;
acquiring the detection confidence of the height limiting device, and judging the reliability of detection according to the numerical value of the detection confidence;
if the detection confidence is judged to be greater than or equal to the confidence threshold, taking the detection result corresponding to the detection confidence as the actual detection result;
if the detection confidence is smaller than the confidence threshold, the model is retrained;
selecting the average precision mean value and the transmission frame number per second as evaluation indexes of a detection algorithm of the height limiting device;
the constructing of the data set of the height limiting device specifically includes:
using a calibrated binocular camera to collect all height limiting devices encountered in a preset road section in the real driving environment of the automobile;
in the acquisition process, the automobile is ensured to run forwards at a preset speed, and all image frames from the imaging of the height limiting device in the binocular camera to the time when the automobile cannot acquire the height limiting data through the height limiting device are acquired;
all image frames are formed into a data set.
2. The height-limiting device detection method according to claim 1, wherein the data set is divided into a training set, a verification set and a test set according to a preset proportion.
3. The method of claim 2, wherein the constructing a data set of the height limiting device further comprises:
marking all height limiting devices in the collected image by using a rectangular frame; wherein the content of the first and second substances,
the marked rectangular frame completely wraps the height limiting part of the height limiting device and does not contain objects except the height limiting device;
marking all height limiting devices contained in the image, and marking a plurality of height limiting devices through rectangular frames in a one-to-one correspondence manner;
a label is embedded for each labeled height limiting device, the label including a category and a name of the height limiting device.
4. The method for detecting a height-limiting device according to claim 3, wherein the training of the features of the height-limiting device in the data set by using the deep learning network to obtain a detection model of the height-limiting device specifically comprises:
the deep learning network is a YOLOV5 network;
inputting the training set of the collected height-limiting data and the labeled corresponding label into a YOLOV5 network for training;
and observing the loss of the training process, and stopping training when the loss of the model is stable so as to obtain the height limiting device detection model.
5. The method for detecting the height-limiting device according to claim 4, wherein the performing error analysis on the detection model of the height-limiting device specifically comprises:
carrying out error analysis according to the input height limiting device test set data;
the average precision mean value and the transmission frame number per second are selected as evaluation indexes of a detection algorithm of the height limiting device, and the calculation formula of the precision is as follows:
accuracy = number of correct positive samples detected by current traversal/number of all positive samples detected by current traversal;
the recall ratio is calculated by the formula:
recall = number of correct positive samples/number of all true values detected by the current traversal.
6. A deep learning based height limiting device detection system, the system comprising:
the data set construction unit is used for constructing a data set of the height limiting device;
the model training unit is used for training the characteristics of the height limiting device in the data set by utilizing a deep learning network to obtain a detection model of the height limiting device;
the model testing unit is used for testing and analyzing errors of the height limiting device detection model to obtain an adjusted height limiting device detection model;
and the result output unit is used for acquiring the height limiting device on the current driving road based on the adjusted height limiting device detection model and outputting early warning information.
7. An intelligent terminal, characterized in that, intelligent terminal includes: the device comprises a data acquisition device, a processor and a memory;
the data acquisition device is used for acquiring data; the memory is to store one or more program instructions; the processor, configured to execute one or more program instructions to perform the method of any of claims 1-5.
8. A computer-readable storage medium having one or more program instructions embodied therein for performing the method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110927771.5A CN113378805B (en) | 2021-08-13 | 2021-08-13 | Height limiting device detection method and system based on deep learning and intelligent terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110927771.5A CN113378805B (en) | 2021-08-13 | 2021-08-13 | Height limiting device detection method and system based on deep learning and intelligent terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113378805A CN113378805A (en) | 2021-09-10 |
| CN113378805B true CN113378805B (en) | 2021-12-17 |
Family
ID=77577220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110927771.5A Active CN113378805B (en) | 2021-08-13 | 2021-08-13 | Height limiting device detection method and system based on deep learning and intelligent terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113378805B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781543B (en) * | 2021-11-10 | 2022-03-18 | 北京中科慧眼科技有限公司 | Binocular camera-based height limiting device detection method and system and intelligent terminal |
| CN114429620A (en) * | 2022-04-07 | 2022-05-03 | 北京中科慧眼科技有限公司 | Training method and system for lower edge position detection model of height limiting device |
| CN114463388A (en) * | 2022-04-13 | 2022-05-10 | 北京中科慧眼科技有限公司 | Binocular camera-based height limiting device detection method and system and intelligent terminal |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20150101072A (en) * | 2014-02-25 | 2015-09-03 | 신도산업 주식회사 | vehicle passing height monitoring system and method |
| CN110069990A (en) * | 2019-03-18 | 2019-07-30 | 北京中科慧眼科技有限公司 | A kind of height-limiting bar detection method, device and automated driving system |
| CN110207650A (en) * | 2019-05-31 | 2019-09-06 | 重庆迪星天科技有限公司 | Automobile-used highway height-limiting frame height measurement method and device based on binocular vision |
| CN111192283A (en) * | 2019-12-26 | 2020-05-22 | 昆山伟宇慧创智能科技有限公司 | Height limiting rod detection and height calculation method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674696B (en) * | 2019-08-28 | 2023-01-13 | 珠海格力电器股份有限公司 | Monitoring method, device, system, monitoring equipment and readable storage medium |
| CN112017165A (en) * | 2020-08-19 | 2020-12-01 | 上海美沃精密仪器股份有限公司 | Lacrimal river height detection method based on deep learning |
-
2021
- 2021-08-13 CN CN202110927771.5A patent/CN113378805B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20150101072A (en) * | 2014-02-25 | 2015-09-03 | 신도산업 주식회사 | vehicle passing height monitoring system and method |
| CN110069990A (en) * | 2019-03-18 | 2019-07-30 | 北京中科慧眼科技有限公司 | A kind of height-limiting bar detection method, device and automated driving system |
| CN110207650A (en) * | 2019-05-31 | 2019-09-06 | 重庆迪星天科技有限公司 | Automobile-used highway height-limiting frame height measurement method and device based on binocular vision |
| CN111192283A (en) * | 2019-12-26 | 2020-05-22 | 昆山伟宇慧创智能科技有限公司 | Height limiting rod detection and height calculation method |
Non-Patent Citations (2)
| Title |
|---|
| A New Real-Time Detection and Tracking Method in Videos for Small Target Traffic Signs;Shaojian Song et al.;《Applied Sciences》;20210330;1-16 * |
| 大件货物远程图像监控及道路通过性离线测量方法的研究;彭琦;《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》;20110915;C034-361 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113378805A (en) | 2021-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113378805B (en) | Height limiting device detection method and system based on deep learning and intelligent terminal | |
| KR101967339B1 (en) | System and Method for Diagnosing Fault and Backup of ADAS Sensors based on Deep Learning | |
| CN104167109B (en) | The detection method of vehicle location and detection device | |
| US20090058622A1 (en) | Method for predicting lane line and lane departure warning system using the same | |
| CN111753649A (en) | Parking space detection method and device, computer equipment and storage medium | |
| CN112163348A (en) | Method and device for detecting road abnormal road surface, simulation method, vehicle and medium | |
| CN114973198A (en) | Course angle prediction method and device of target vehicle, electronic equipment and storage medium | |
| CN113781543B (en) | Binocular camera-based height limiting device detection method and system and intelligent terminal | |
| CN109580632B (en) | Defect determination method, device and storage medium | |
| CN116702000B (en) | Road surface quality dynamic monitoring and evaluating method based on multi-layer data fusion | |
| CN112183382A (en) | Unmanned traffic light detection and classification method and device | |
| CN111754574A (en) | Distance testing method, device and system based on binocular camera and storage medium | |
| CN113870754B (en) | Method and system for judging defects of panel detection electronic signals | |
| CN114062656B (en) | Geological disaster prediction method and system | |
| CN115909271A (en) | Parking space identification method and device, vehicle and storage medium | |
| CN114463388A (en) | Binocular camera-based height limiting device detection method and system and intelligent terminal | |
| CN112733778B (en) | Vehicle front guide determination method and device and computer equipment | |
| WO2022041971A1 (en) | Positioning reliability test method and related device | |
| CN114429620A (en) | Training method and system for lower edge position detection model of height limiting device | |
| CN111539279B (en) | Road height limit detection method, device, equipment and storage medium | |
| CN114724107A (en) | Image detection method, device, equipment and medium | |
| CN113419075B (en) | Ship speed measuring method, system, device and medium based on binocular vision | |
| CN114677574B (en) | Method and system for diagnosing image fault for automatic driving | |
| CN116503489A (en) | Camera calibration file self-diagnosis method and device, electronic equipment and storage medium | |
| CN116168087A (en) | Verification method and device for road side camera calibration result and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |