CN114550139A - Lane line detection method and device - Google Patents
Lane line detection method and device Download PDFInfo
- Publication number
- CN114550139A CN114550139A CN202210199343.XA CN202210199343A CN114550139A CN 114550139 A CN114550139 A CN 114550139A CN 202210199343 A CN202210199343 A CN 202210199343A CN 114550139 A CN114550139 A CN 114550139A
- Authority
- CN
- China
- Prior art keywords
- lane line
- line detection
- image
- layer
- 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.)
- Pending
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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and a device for detecting a lane line, which are used for solving the defects that in the prior art, a road image detected by a vehicle is fuzzy, a real lane line is easy to be shielded, and is easy to be interfered by light rays, the missing detection rate is high, the detection error is large, and the anti-interference capability is poor, realizing the improvement of the anti-interference capability, reducing the missing detection rate and the detection error, reducing the calculation amount and improving the calculation speed. According to the lane line detection method and device provided by the invention, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing detection rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a lane line detection method and a lane line detection device.
Background
With the rapid development of artificial intelligence technology, various scenes such as traffic, logistics, security inspection and the like are applied to the artificial intelligence technology, for example, in an unmanned or assisted driving scene, lane line detection is needed, and the lane line detection is a technology for realizing automatic identification and perception detection of a traffic indication line on a road by a vehicle.
At present, in a scene of lane line detection, road images detected by vehicles are fuzzy, real lane lines are easy to be shielded and are easy to be interfered by light rays, the missing detection rate is high, the detection error is large, and the anti-interference capability is poor.
Disclosure of Invention
The invention provides a method and a device for detecting a lane line, which are used for solving the defects that in the prior art, a road image detected by a vehicle is fuzzy, a real lane line is easy to be shielded, and is easy to be interfered by light rays, the missing detection rate is high, the detection error is large, and the anti-interference capability is poor, realizing the improvement of the anti-interference capability, reducing the missing detection rate and the detection error, reducing the calculation amount and improving the calculation speed.
The invention provides a lane line detection method, which comprises the following steps: acquiring an image to be detected; inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model; wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for obtaining global information based on the compression features, and the grid classification layer is used for obtaining lane line detection results based on the compression features and the global information.
According to the lane line detection method provided by the invention, the training process of the lane line detection model comprises the following steps: acquiring an original sample image and a random-shaped blocking block; determining an occlusion sample image based on the original sample image and the random-shape occlusion block; and training the lane line detection model based on the original sample image and the occlusion sample image.
According to the lane line detection method provided by the invention, the training process of the lane line detection model comprises the following steps: obtaining a plurality of sample images and lane line sample data corresponding to the sample images; and training the lane line detection model by taking the sample image as a sample and taking lane line sample data corresponding to the sample image as a sample label.
According to the lane line detection method provided by the invention, the training process of the lane line detection model comprises the following steps: and training the lane line detection model based on a semantic segmentation algorithm.
According to the lane line detection method provided by the invention, the lane line detection model further comprises: and the characteristic extraction layer is used for extracting various optional characteristics based on the image to be detected.
According to the lane line detection method provided by the invention, the lane line detection model further comprises: the feature pyramid network layer is used for obtaining a plurality of multi-scale representation features based on the plurality of candidate features, and the feature compression layer is used for extracting a plurality of compression features based on the plurality of multi-scale representation features.
According to the lane line detection method provided by the invention, the lane line detection method further comprises the following steps: and mapping the lane line detection result to the image to be detected to obtain the coordinate information of the lane line on the image to be detected.
The present invention also provides a lane line detection apparatus, including: the acquisition module is used for acquiring an image to be detected; the detection module is used for inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model; wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for obtaining global information based on the compression features, and the grid classification layer is used for obtaining lane line detection results based on the compression features and the global information.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of the lane line detection method.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the lane line detection method as described in any one of the above.
The invention also provides a computer program product comprising a computer program which, when executed by a processor, carries out the steps of the lane line detection method as described in any one of the above.
According to the lane line detection method and device provided by the invention, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing detection rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a lane line detection method provided by the present invention;
FIG. 2 is a schematic structural diagram of a lane marking detection apparatus according to the present invention;
fig. 3 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. 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 following describes the lane marking detection method and apparatus of the present invention with reference to fig. 1 to 3.
As shown in fig. 1, the present invention provides a lane line detection method, including: as follows from step 110 to step 120.
And step 110, acquiring an image to be detected.
It can be understood that the image to be detected is a road image of a lane line to be detected, the image to be detected can be a road image shot by a camera of a vehicle, and the image to be detected can include the lane line, the vehicle, a pedestrian, an animal, a transportation facility or other objects.
It is worth mentioning that in the image to be detected, the real lane line may be partially blocked by the vehicle, the pedestrian, the animal or other objects, and there is interference in detecting the lane line when the lane line is partially blocked.
And inputting the image to be detected into the lane line detection model to obtain a lane line detection result output by the lane line detection model.
Wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for acquiring global information based on the compression features, and the grid classification layer is used for acquiring a lane line detection result based on the compression features and the global information.
It will be appreciated that the lane line detection model may be a neural network model, such as a convolutional neural network or a residual neural network.
The lane line detection model can comprise a feature compression layer, a full connection layer and a grid classification layer, wherein the output end of the feature compression layer is connected with the input end of the full connection layer, and the output end of the feature compression layer and the output end of the full connection layer are connected with the input end of the grid classification layer.
The feature Compression layer may perform fcm (feature Compression motion) operation on a feature with stride of 16 in the image to be detected. The feature compression layer can be used to maintain the height of the features, compressing across the width. Because the lane lines appear as stripes in the image and mostly start from the left, right or bottom of the image, gather and disappear in front. The feature compression layer is more beneficial to extracting the features of the shape. By adopting the characteristic compression layer, the lane line detection model is simpler, the calculated amount is smaller, and the method is more suitable for being deployed at a mobile terminal.
The full connection layer can carry out the full connection operation to the compression characteristic that the characteristic compression layer extracted to acquire global information based on a plurality of compression characteristics, because the lane line all has certain shape prior in shape and direction, to the camera visual angle in the car, the lane line is slenderness type, directional place ahead to disappear in place far away. The global information is acquired by utilizing the full-connection layer, so that the whole trend of the lane line can be conveniently deduced based on the global information after partial lane line positions are extracted from the local area, the whole can be deduced through parts, and the calculation amount is reduced.
The grid classification layer can obtain grid classification results of different types of lane lines according to the compression characteristics and the global information reshape, the results are finally output in a grid mode, lane line detection results can be displayed in a grid mode, each grid can have a score, and the scores can be the probability that the grid is a lane line.
In some embodiments, the training process of the lane line detection model includes: acquiring a plurality of sample images and lane line sample data corresponding to the sample images; and training the lane line detection model by taking the sample image as a sample and taking the lane line sample data corresponding to the sample image as a sample label.
The lane line detection model can be trained through a large number of sample images and lane line sample data marked in advance, and the training accuracy of the lane line detection model can be ensured in a supervised learning mode.
The deep learning neural network used by the lane line detection model can pick out the features in the input sample image, each feature is used for obtaining an output result, each output result is compared with the sample label, the features meeting the requirements through comparison can be reserved, the features meeting the requirements through comparison are ignored through Loss parameters, the core features needing to be memorized can be finally learned through continuous iterative training of a large number of input sample images, different core features are classified, and the newly input image to be detected can be finally distinguished according to the core features.
Before the lane line detection model is trained, a filter of a convolution layer of the deep learning neural network is completely random, and the filter can not activate any feature, namely can not detect any feature, in the training process, the blank filter is modified in weight to enable the blank filter to detect a specific scene, and the method is a supervised learning mode.
It is worth mentioning that lane line detection is a fundamental module in unmanned driving. Along with the development of economy, building pavements are more and more, and lane line scenes are more and more diversified. In the early stage, a plurality of implementations based on the traditional image processing methods exist, such as line detection by canny and sobel, feature extraction by a structure tensor and a strip filter, post-processing by hough, ransac, Kalman filtering and the like. However, the robustness is not strong, especially when the conditions of blurring, being influenced by illumination or being shielded occur, the conditions of missing detection or false detection often occur, and the traditional algorithm for the night scene cannot detect the condition. With deep learning and the rapid development of big data, more and more researchers are turning to this direction.
The inventor finds in research that the task can be regarded as a segmentation task, each pixel point in an image needs to be classified, but the problem of low speed exists, most of lane line detection runs at the end side, and strict requirements are imposed on the speed of an algorithm.
In addition, the segmentation normally adopts a full convolution mode to obtain a segmentation result, and the receptive field of each pixel of the full convolution is almost local. The positioning of the lane lines is closely related to the global information of the surrounding traffic flow, and particularly, the global information is more important when the lane lines are shielded and fuzzy. How to improve the performance of the algorithm under the condition of ensuring the speed of the algorithm, and considering the prior information and the global information of the lane line becomes a key problem.
According to the lane line detection method provided by the invention, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing detection rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
In some embodiments, the training process of the lane line detection model includes: acquiring an original sample image and a random-shaped blocking block; determining an occlusion sample image based on the original sample image and the random-shape occlusion block; and training the lane line detection model based on the original sample image and the shielding sample image.
It can be understood that, in the course of training the lane line detection model, a random-shaped blocking block may be generated, the shape of the random-shaped blocking block is random, and may be regular round, square or trapezoid, or may be irregular, the number of the random-shaped blocking blocks may be one or multiple, the random-shaped blocking block may be covered on the original sample image, the random-shaped blocking block may block a partial region on the original sample image, the blocked content may be a vehicle, a pedestrian or a partial lane line region, where the blocked region of the random-shaped blocking block is not limited, as long as the partial blocking is performed, the blocked sample image is obtained.
The shielding sample images and the original sample images can be used for training the lane line detection model together, so that the number of training samples can be increased, meanwhile, the shielding sample images are used for training the lane line detection model, the situation that the part of the shielding part of the image to be detected is easy to appear under the practical application scene can be simulated, the fault tolerance of the trained lane line detection model is stronger, and the anti-interference capability is improved.
In some embodiments, the training process of the lane line detection model includes: and training the lane line detection model based on a semantic segmentation algorithm.
It is understood that in the field of image recognition and computer vision, image segmentation refers to dividing an image into a plurality of mutually disjoint regions according to features such as gray scale, color, spatial texture, geometric shape, etc., so that the features show consistency or similarity in the same region and obviously differ from region to region. In other words, the common segmentation is to separate pixel regions of objects belonging to different categories, such as a foreground and a background, and separate a region where a cat or a dog is located from a background region, and the semantic segmentation is to classify the semantics of each region, that is, what object the region is based on the common segmentation, so that all the objects in the picture can be indicated by their respective categories, and here, a semantic segmentation algorithm can be used to train the lane line detection model, so that the lane line detection model can acquire the capability of accurately segmenting the image to be detected.
In some embodiments, the lane line detection model further comprises: and the characteristic extraction layer is used for extracting various alternative characteristics based on the image to be detected.
It is understood that the feature extraction layer backbone can extract various candidate features from the image to be detected according to the requirements of speed and performance, such as res18, res34, res50, mobilenetv2, mobilenetv3, mobillenext, and the like.
In some embodiments, the lane line detection model further comprises: the feature pyramid network layer is used for obtaining a plurality of multi-scale representation features based on a plurality of optional features, and the feature compression layer is used for extracting a plurality of compression features based on the plurality of multi-scale representation features.
It can be understood that Feature Pyramid Network (Feature Pyramid Network) is abbreviated as FPN, and like TDM (Top-Down Modulation) method, FPN is a Top-Down Feature fusion method, but FPN is a multi-scale target detection algorithm, i.e. there is not only one Feature prediction layer.
The network structure of the feature pyramid network can be used for fusing a feature map with strong low-resolution semantic information and a feature map with weak high-resolution semantic information and rich spatial information on the premise of increasing less calculation amount. The feature pyramid network can use a nearest neighbor value interpolation method, and can reserve the semantic information of the feature map to the maximum extent in the up-sampling process, so that the feature map is fused with the corresponding feature map with abundant spatial information in the bottom-up process, and the feature map with good spatial information and stronger semantic information is obtained.
Although some algorithms also adopt multi-scale feature fusion to perform target detection, the method usually only uses one scale feature obtained after fusion, although the method can use semantic information of top-level features and detail information of bottom-level features, some deviation can be caused in the processes of feature deconvolution and the like, and the detection precision can be adversely affected by only using the feature obtained after fusion to perform prediction.
Starting from the problem, the FPN method can predict fusion characteristics of a plurality of different scales, and maximize detection precision.
The idea of the FPN characteristic pyramid is derived from multi-scale identification in the traditional algorithm, and the specific operation is that an original image is zoomed to a state with different scales, the zoomed image is applied to the global characteristics of the image, and the zoomed image is applied to the detail characteristics. The deeper the feature map of the deep learning network, the more global and abstract features are possessed. In the image classification task, the deep-level features keep good translation invariance, and no matter where the classified object image is, the deep-level global features can still obtain information.
However, in the field of image recognition, translation invariance in image classification does not hold, and the position of an object is recognized not only for recognizing the classification of the object, so how to combine shallow and deep information is an important problem.
The FPN solves the problem of shallow and deep layer combination by a mode of longitudinal addition of transverse connecting sections. The overall structure adopts a top-down information structure, and a plurality of multi-scale representation features can be obtained based on a plurality of alternative features, so that a plurality of multi-scale representation features can be provided for a feature compression layer, and the feature compression layer is used for extracting a plurality of compression features based on the plurality of multi-scale representation features.
The feature pyramid network layer can promote multi-scale feature representation.
In some embodiments, the lane line detection method further includes: and mapping the lane line detection result to the image to be detected to obtain the coordinate information of the lane line on the image to be detected.
It can be understood that the lane line detection result may be presented in the form of grid grids, each grid may have a score, and the score may be a probability that the grid is a lane line, where the score of each output grid may be used as a weight for each type of lane line, and the id of the grid is an amount to be weighted, so as to find the floating point coordinates of the lane line on the grid. And mapping the coordinates to the size of the image to be detected to obtain the coordinates of each type of lane line on the image to be detected.
The following describes the lane line detection device provided by the present invention, and the lane line detection device described below and the lane line detection method described above may be referred to in correspondence with each other.
As shown in fig. 2, the present invention also provides a lane line detecting apparatus, including: an acquisition module 210 and a detection module 220.
The obtaining module 210 is configured to obtain an image to be detected.
The detection module 220 is configured to input the image to be detected into the lane line detection model to obtain a lane line detection result output by the lane line detection model.
Wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for acquiring global information based on the compression features, and the grid classification layer is used for acquiring a lane line detection result based on the compression features and the global information.
According to the lane line detection device, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
According to the lane line detection device provided by the invention, the training process of the lane line detection model comprises the following steps: acquiring an original sample image and a random-shaped blocking block; determining an occlusion sample image based on the original sample image and the random-shape occlusion block; and training a lane line detection model based on the original sample image and the shielding sample image.
According to the lane line detection device provided by the invention, the training process of the lane line detection model comprises the following steps: obtaining a plurality of sample images and lane line sample data corresponding to the sample images; and training the lane line detection model by taking the sample image as a sample and taking lane line sample data corresponding to the sample image as a sample label.
According to the lane line detection device provided by the invention, the training process of the lane line detection model comprises the following steps: and training the lane line detection model based on a semantic segmentation algorithm.
According to a lane line detection apparatus provided by the present invention, the lane line detection model further includes: and the characteristic extraction layer is used for extracting various optional characteristics based on the image to be detected.
According to a lane line detection apparatus provided by the present invention, the lane line detection model further includes: the feature pyramid network layer is used for obtaining a plurality of multi-scale representation features based on the plurality of candidate features, and the feature compression layer is used for extracting a plurality of compression features based on the plurality of multi-scale representation features.
According to a lane line detection apparatus provided by the present invention, the lane line detection method further includes: and mapping the lane line detection result to the image to be detected to obtain the coordinate information of the lane line on the image to be detected.
Fig. 3 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 3: a processor (processor)310, a communication Interface (communication Interface)320, a memory (memory)330 and a communication bus 340, wherein the processor 310, the communication Interface 320 and the memory 330 communicate with each other via the communication bus 340. The processor 310 may invoke logic instructions in the memory 330 to perform a lane line detection method comprising: acquiring an image to be detected; inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model; wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for acquiring global information based on the compression features, and the grid classification layer is used for acquiring a lane line detection result based on the compression features and the global information.
According to the electronic equipment provided by the invention, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing detection rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
In addition, the logic instructions in the memory 330 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being storable on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer being capable of executing the lane line detection method provided by the above methods, the method including: acquiring an image to be detected; inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model; wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for acquiring global information based on the compression features, and the grid classification layer is used for acquiring a lane line detection result based on the compression features and the global information.
According to the computer program product provided by the invention, the lane detection result is obtained from the image to be detected through the lane detection model consisting of the characteristic compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing detection rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements a lane line detection method provided by the above methods, the method including: acquiring an image to be detected; inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model; wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for acquiring global information based on the compression features, and the grid classification layer is used for acquiring a lane line detection result based on the compression features and the global information.
According to the non-transitory computer readable storage medium provided by the invention, the lane line detection result is obtained from the image to be detected through the lane line detection model consisting of the feature compression layer, the full connection layer and the grid classification layer, so that the anti-interference capability can be improved, the missing rate and the detection error can be reduced, the calculated amount can be reduced, and the calculation speed can be increased.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A lane line detection method is characterized by comprising the following steps:
acquiring an image to be detected;
inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model;
wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for obtaining global information based on the compression features, and the grid classification layer is used for obtaining lane line detection results based on the compression features and the global information.
2. The lane line detection method according to claim 1, wherein the training process of the lane line detection model includes:
acquiring an original sample image and a random-shaped blocking block;
determining an occlusion sample image based on the original sample image and the random-shape occlusion block;
and training the lane line detection model based on the original sample image and the occlusion sample image.
3. The lane line detection method according to claim 1, wherein the training process of the lane line detection model includes:
obtaining a plurality of sample images and lane line sample data corresponding to the sample images;
and training the lane line detection model by taking the sample image as a sample and taking lane line sample data corresponding to the sample image as a sample label.
4. The lane line detection method according to claim 1, wherein the training process of the lane line detection model includes:
and training the lane line detection model based on a semantic segmentation algorithm.
5. The lane line detection method according to any one of claims 1 to 4, wherein the lane line detection model further includes:
and the characteristic extraction layer is used for extracting various optional characteristics based on the image to be detected.
6. The lane line detection method according to claim 5, wherein the lane line detection model further comprises:
the feature pyramid network layer is used for obtaining a plurality of multi-scale representation features based on the plurality of candidate features, and the feature compression layer is used for extracting a plurality of compression features based on the plurality of multi-scale representation features.
7. The lane line detection method according to any one of claims 1 to 4, further comprising:
and mapping the lane line detection result to the image to be detected to obtain the coordinate information of the lane line on the image to be detected.
8. A lane line detection apparatus, comprising:
the acquisition module is used for acquiring an image to be detected;
the detection module is used for inputting the image to be detected into a lane line detection model to obtain a lane line detection result output by the lane line detection model;
wherein, lane line detection model includes: the system comprises a feature compression layer, a full connection layer and a grid classification layer, wherein the feature compression layer is used for extracting a plurality of compression features based on an image to be detected, the full connection layer is used for obtaining global information based on the compression features, and the grid classification layer is used for obtaining lane line detection results based on the compression features and the global information.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the lane line detection method according to any one of claims 1 to 7 are implemented when the processor executes the program.
10. A non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor implements the steps of the lane line detection method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210199343.XA CN114550139A (en) | 2022-03-02 | 2022-03-02 | Lane line detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210199343.XA CN114550139A (en) | 2022-03-02 | 2022-03-02 | Lane line detection method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114550139A true CN114550139A (en) | 2022-05-27 |
Family
ID=81662648
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210199343.XA Pending CN114550139A (en) | 2022-03-02 | 2022-03-02 | Lane line detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114550139A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115273013A (en) * | 2022-09-27 | 2022-11-01 | 江西小马机器人有限公司 | Lane line detection method, system, computer and readable storage medium |
-
2022
- 2022-03-02 CN CN202210199343.XA patent/CN114550139A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115273013A (en) * | 2022-09-27 | 2022-11-01 | 江西小马机器人有限公司 | Lane line detection method, system, computer and readable storage medium |
| CN115273013B (en) * | 2022-09-27 | 2024-05-03 | 江西小马机器人有限公司 | Lane line detection method, system, computer and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111666921B (en) | Vehicle control method, apparatus, computer device, and computer-readable storage medium | |
| CN112308826B (en) | Bridge structure surface defect detection method based on convolutional neural network | |
| CN102314599A (en) | Identification and deviation-detection method for lane | |
| CN110163109B (en) | Lane line marking method and device | |
| CN110008900B (en) | Method for extracting candidate target from visible light remote sensing image from region to target | |
| CN104915642B (en) | Front vehicles distance measuring method and device | |
| CN112666553B (en) | Road ponding identification method and equipment based on millimeter wave radar | |
| CN113240623B (en) | Pavement disease detection method and device | |
| CN114841910A (en) | Vehicle-mounted lens shielding identification method and device | |
| CN111213154A (en) | Lane line detection method, lane line detection equipment, mobile platform and storage medium | |
| Magnussen et al. | A survey of the inadequacies in traffic sign recognition systems for autonomous vehicles | |
| CN114743181A (en) | Road vehicle target detection method and system, electronic device and storage medium | |
| CN115457358A (en) | Image and point cloud fusion processing method and device and unmanned vehicle | |
| Parvin et al. | Vehicle number plate detection and recognition techniques: a review | |
| Ghahremannezhad et al. | Automatic road detection in traffic videos | |
| CN114550139A (en) | Lane line detection method and device | |
| Santos et al. | Learning to segment roads for traffic analysis in urban images | |
| Arthi et al. | Object detection of autonomous vehicles under adverse weather conditions | |
| Zuo et al. | Road model prediction based unstructured road detection | |
| CN116664851A (en) | Automatic driving data extraction method based on artificial intelligence | |
| Lim et al. | Vision-based recognition of road regulation for intelligent vehicle | |
| Ballinas-Hernández et al. | Marked and unmarked speed bump detection for autonomous vehicles using stereo vision | |
| Ab Ghani et al. | Lane Detection Using Deep Learning for Rainy Conditions | |
| CN115100650A (en) | Expressway abnormal scene denoising and identifying method and device based on multiple Gaussian models | |
| CN114092880A (en) | Airport runway large particle foreign matter detection method based on video analysis |
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 |