CN115457277A - Intelligent pavement disease identification and detection method and system - Google Patents
Intelligent pavement disease identification and detection method and system Download PDFInfo
- Publication number
- CN115457277A CN115457277A CN202211165084.5A CN202211165084A CN115457277A CN 115457277 A CN115457277 A CN 115457277A CN 202211165084 A CN202211165084 A CN 202211165084A CN 115457277 A CN115457277 A CN 115457277A
- Authority
- CN
- China
- Prior art keywords
- detection
- image
- road
- module
- vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- 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
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
Abstract
The invention discloses an intelligent pavement disease identification and detection method and system, wherein a motion camera and a GPS and Beidou positioning module are adopted as acquisition equipment, an improved Kalman filtering algorithm, a Crack-QuickSort algorithm and an improved Crack-wut algorithm are adopted to carry out preprocessing and processing operation on positioning data and image data, the position, length, width and area information of a pavement disease can be accurately solved, and the pavement surface damage state index PCI is obtained by combining the information. The system comprises a vehicle position acquisition module, a road image detection module, a road image feature extraction module and a pavement defect report output module. The invention has the advantages of realizing high-precision pavement disease detection, constructing an integral pavement disease detection system, reducing the cost of pavement detection equipment by adopting a method of recognizing diseases by using a foreground camera and increasing the intelligent degree in the whole detection process.
Description
Technical Field
The invention belongs to the technical field of road engineering detection, and particularly relates to a pavement disease identification and detection method and system.
Background
In the research of high-precision detection of road diseases, a disease detection method based on deep learning is widely developed, and the high-precision detection of the diseases usually comprises three parts, namely real-time positioning of the diseases, high-precision segmentation of the diseases and main feature extraction of the diseases. These three parts are mainly realized by the following ways:
(1) For the real-time positioning research of diseases, researchers adopt YOLO and Deepsort algorithms to realize real-time detection and tracking of pavement diseases, a detector and a tracker are constructed, gamma transformation is adopted in the training process to simulate a backlight image, training data are enhanced, and meanwhile, the detection performance of a model in a backlight environment is improved. On the other hand, on the tracker, the VGG16 full convolution part is used as a depth feature extraction network, and good guarantee is obtained on the calculation rate and the effect. But the defects still exist, the main defects are that the whole process of disease detection cannot be completed only by positioning, the precision of a disease anchor frame is insufficient, the capability of treating special pavement diseases is insufficient, and the like.
(2) For the high-precision segmentation research of diseases, a U-Net neural network is adopted, affine transformation is adopted to perform data enhancement processing on rotation operation in the training process of a segmentation model, and a data set is expanded. And finally, importing the segmentation result into a characteristic statistics module, judging the crack characteristics, identifying the crack degree and providing data support for index calculation. The task segmentation difficulty is reduced, and the applicability of the trained model is improved. But the method still has the defects of insufficient precision, incapability of extracting disease characteristics end to end, lack of segmentation capability in a complex environment and incapability of finishing the whole disease detection process due to insufficient systematicness.
(3) For the extraction process of main disease features, researchers often combine some data image processing means to perform post-processing on a high-precision segmented crack image, for example, the shortest distance is adopted to define the crack length, and meanwhile, the average crack width is calculated according to the crack area.
Disclosure of Invention
The invention aims to: the method and the system for identifying and detecting the pavement diseases are provided to solve the problems in the prior art. In order to achieve the purpose, the invention provides the following technical scheme: a pavement disease identification and detection method comprises the following steps:
s1, obtaining the position of a vehicle: calculating to obtain the optimal position information of the vehicle according to the current position and speed information of the vehicle;
s2, detecting a road image: obtaining an interested area of a corresponding road image based on road image data of the current optimal vehicle position information;
s3, extracting road image features: obtaining the characteristics of the corresponding road surface disease image based on the interested area of the road image;
s4, outputting a pavement disease monitoring report form: and outputting a road disease detection report by taking the position and speed information of the vehicle and the characteristics of the corresponding road disease image as input.
Further, in the step S1, the speed of the vehicle is subjected to smoothing filtering processing based on the current vehicle position and speed information, and an error of the vehicle position information is corrected, so that the optimal vehicle position information is obtained.
Further, in the above step S2, based on the road image data of the current vehicle position information, the acceleration sensor acquires acceleration data in the vertical direction of the vehicle, and the image position is corrected.
Further, in the foregoing step S2, an improved Crack-QuickSort model is adopted to extract the region of interest of the road image, where the improved Crack-QuickSort model is composed of a target detection module and a target tracking module; the method comprises the steps that four serial CrackInception blocks are added in an improved CrackQuickSort model feature extraction layer, three full convolution layers are removed, and each CrackInception block comprises three parallel Inception-v3 structures and is used for extracting an interested area of a road image. In the target tracking module, detection frames with the same tracking id are recorded, an image registration algorithm of a longitudinal crack detection frame based on key corner detection is added, image matching is carried out on the longitudinal cracks of the road, and complete detection information of a single longitudinal crack is obtained, wherein the key corner detection adopts an SIFT corner detection algorithm.
Further, in the step S3, the features of the road surface disease image are extracted by using an improved Crack-Wnet algorithm, specifically: adding a super-feature extraction layer between a U-Net neural network coding layer and a decoding layer, wherein the super-feature extraction layer comprises a Dense-block and a W-block which are connected in parallel, the Dense-block comprises two fully-connected layers of 2048 units connected in series, each node in the W-block comprises three convolution layers, the convolution adopts a cavity convolution with the size of 3 x 3, the convolution core expansion rates of the three convolution layers are 1,2,3 respectively, and the super-feature extraction layer is used for enhancing the depth feature extraction capability of road surface disease images.
On the other hand, the invention provides an intelligent pavement disease identification and detection system, which comprises:
the vehicle position acquisition module is based on a positioning system on a vehicle, takes the current vehicle position and speed information as input, and takes the optimal vehicle position information as output;
the road image detection module is based on a shooting device on the vehicle, takes the road image data of the current vehicle position information as input, and takes the interested area of the corresponding road image as output;
the road image feature extraction module takes the region of interest of the road image as input and the feature of the corresponding road surface disease image as output;
and the road surface disease report output module is used for outputting a disease detection report, and the disease detection report is output by taking the vehicle position and speed information and the characteristics of the corresponding road surface disease image as input.
Further, the aforementioned vehicle position acquisition module is configured to: inputting the vehicle position information into a vehicle position acquisition module, performing smooth filtering processing on the speed of the vehicle, correcting the error of the position information, and acquiring the optimal vehicle position information
Further, the aforementioned road image detection module is configured to: road image data of current vehicle position information are input to a road image detection module, acceleration data in the vertical direction of the vehicle are obtained by an acceleration sensor, and the position of the image is corrected.
Further, the road image detection module comprises an improved Crack-QuickSort model formed by a target detection module and a target tracking module; four serial CrackInception blocks are added in an improved CrackQuickSort model feature extraction layer, three full convolution layers are removed, and each CrackInception block comprises three parallel Inception-v3 structures and is used for extracting an interested area of a road image.
In the target tracking module, detection frames with the same tracking id are recorded, an image registration algorithm of a longitudinal crack detection frame based on key corner detection is added, image matching is carried out on the longitudinal cracks of the road, and complete detection information of a single longitudinal crack is obtained, wherein the key corner detection adopts an SIFT corner detection algorithm.
Further, the road image feature extraction module adds a super-feature extraction layer between the U-Net neural network coding layer and the decoding layer by using an improved Crack-wnt algorithm, the super-feature extraction layer is a Dense-block and a W-block which are connected in parallel, the Dense-block is composed of two full-connection layers of 2048 units which are connected in series, each node in the W-block is composed of three convolution layers, the convolution adopts a cavity convolution with the size of 3 x 3, the convolution core expansion rates of the three convolution layers are 1,2,3 respectively, and the super-feature extraction layer is used for enhancing the depth feature extraction capability of the road surface disease image.
Compared with the prior art, the invention has the following beneficial effects: the rapid detection and statistics functions of diseases are realized by adopting a mode of combining a target detector and a tracker; the image in the detection frame is adopted for semantic segmentation, so that the accuracy and performance of disease segmentation are improved; the road is segmented, and the real-time calculation of the road damage index is realized by combining the semantic segmentation result, so that the actual output result has actual engineering value. In addition, the method for recognizing the diseases by adopting the foreground camera reduces the cost of the pavement detection equipment and increases the intelligent degree in the whole detection process.
Drawings
FIG. 1 is a flow chart of the algorithm of the present invention.
Fig. 2 is a layout diagram of equipment of an intelligent pavement damage recognition and detection system.
FIG. 3 is a flow chart of the Crack-QuickSort model detection of the present invention.
Detailed Description
In order to better understand the technical content of the present invention, specific embodiments are described below with reference to the accompanying drawings.
Aspects of the invention are described herein with reference to the accompanying drawings, in which a number of illustrative embodiments are shown. Embodiments of the invention are not limited to those described in the figures. It is to be understood that the invention is capable of implementation in any of the numerous concepts and embodiments described hereinabove or described in the following detailed description, since the disclosed concepts and embodiments are not limited to any embodiment. In addition, some aspects of the present disclosure may be used alone, or in any suitable combination with other aspects of the present disclosure.
As shown in fig. 1, the algorithm flow chart of the invention, an intelligent pavement disease identification and detection method, comprises the steps of:
s1, obtaining the position of a vehicle: calculating to obtain the optimal position information of the vehicle according to the current position and speed information of the vehicle; based on the current vehicle position and speed information, obtaining the optimal vehicle position information by using a Kalman filtering algorithm, wherein the Kalman filtering algorithm specifically comprises the following steps: the speed of the vehicle is subjected to smoothing filtering processing, and the error of the vehicle position information is corrected.
S2, detecting a road image: obtaining an interested area of a corresponding road image based on road image data of the current optimal vehicle position information; the method specifically comprises the following steps: and acquiring acceleration data of the vehicle in the vertical direction by using an acceleration sensor, and correcting the position of the image. Extracting an interested area of the road image by adopting an improved Crack-QuickSort model, wherein the improved Crack-QuickSort model consists of a target detection module and a target tracking module; wherein:
four serial CrackInclusion blocks are added in an improved CrackQuickSort model feature extraction layer, three full convolution layers are removed, and each CrackInclusion block comprises three parallel Inclusion-v 3 structures and is used for extracting an interested area of a road image. The method comprises the following steps of utilizing an improved Sort algorithm in the target tracking module for carrying out image matching on the longitudinal cracks of the road during image detection to obtain complete detection information of a single longitudinal crack, wherein the improved Sort algorithm specifically comprises the following steps: in the tracking process, detecting frames with the same tracking id are recorded, and a longitudinal crack detecting frame image registration algorithm based on key corner detection is added, wherein the key corner detection adopts an SIFT corner detection algorithm.
S3, extracting road image features: and obtaining the characteristics of the corresponding road surface disease image based on the interested area of the road image. The method specifically comprises the following steps: the method for extracting the features of the pavement disease image by using the improved Crack-Wnet algorithm comprises the following steps: adding a super-feature extraction layer between a U-Net neural network coding layer and a decoding layer, wherein the super-feature extraction layer comprises a Dense-block and a W-block which are connected in parallel, the Dense-block comprises two fully-connected layers of 2048 units connected in series, each node in the W-block comprises three convolution layers, the convolution adopts a cavity convolution with the size of 3 x 3, the convolution core expansion rates of the three convolution layers are 1,2,3 respectively, and the super-feature extraction layer is used for enhancing the depth feature extraction capability of road surface disease images.
S4, outputting a pavement disease monitoring report form: and outputting a road disease detection report by taking the position and speed information of the vehicle and the characteristics of the corresponding road disease image as input.
As shown in FIG. 2, a GPS and Beidou positioning module antenna is arranged at the top of the vehicle to collect position and speed information of the vehicle. A gopro motion camera is arranged in front of the vehicle, the inclination angle is 42 degrees, and road image data when the vehicle runs are collected.
In a vehicle position acquisition module, the vehicle position and speed information is calculated by using an improved Kalman filtering algorithm to obtain the optimal vehicle position information, and as shown in a table 1 for computing the optimal vehicle position information, the accuracy in the table is improved into the error reduction rate between the positioning longitude and latitude and the actual longitude and latitude of the road in the map.
TABLE 1
And the road image detection module is used for taking road image data of the position information of one kilometer of the current vehicle as input and an interested area of the corresponding road image as output based on the vehicle position information under the optimal estimation output by the vehicle position acquisition module. The road image detection module comprises a target detection module and a target tracking module.
As shown in fig. 3, the road image detection module includes an improved Crack-QuickSort model composed of a target detection module and a target tracking module; four serial CrackInception blocks are added in a characteristic extraction layer, three full convolution layers are removed, and each CrackInception block comprises three parallel Inception-v3 structures and is used for extracting an interested area of a road image; the method comprises the steps of inputting road image data within one kilometer into a Crack-quick Sort model to obtain a tracked image segmentation interesting region, detecting a disease target through a target detection module, tracking through a target tracking module to finally obtain the interesting region, wherein the tracked interesting region is shown in an image segmentation interesting region output result table in a table 2, and relevant values of the interesting region in the table are normalized according to image sizes.
TABLE 2
The road image feature extraction module is characterized in that a super feature extraction layer is added between a U-Net neural network coding layer and a decoding layer by using an improved Crack-wnt algorithm, the super feature extraction layer is a Dense-block and a W-block which are connected in parallel, the Dense-block is composed of two full-connection layers of 2048 units in series, each node in the W-block is composed of three convolution layers, the convolution adopts a cavity convolution with the size of 3 x 3, the convolution core expansion rates of the three convolution layers are 1,2,3 respectively, and the super feature extraction module is used for enhancing the depth feature extraction capability of a road surface disease image. The coordinates of the image segmentation interesting region are mapped through a perspective transformation matrix, the mapped coordinates are input into a road image feature extraction module, the length, width and area information of the road surface diseases are obtained, the perspective transformation matrix is obtained by the camera inclination angle, and the disease feature results obtained through calculation are shown in a table 3.
TABLE 3
| Disease category | Length m | Width m | Area/area of influence m 2 |
| Pit slot | 0.01 | 0.02 | 0.0002 |
| Network crack | 0.05 | 0.04 | 0.002 |
| Transverse crack | 0.05 | 0.02 | 0.001 |
The vehicle position information output by the vehicle position acquisition module and the disease length, width and area information output by the road image feature extraction module are input into the road surface disease report output module for sorting, calculating and summarizing, and a road surface disease detection result report within one kilometer is obtained as shown in table 4, wherein a road surface damage state index (PCI) is calculated by combining with a road technical condition evaluation standard.
TABLE 4
And the PCI index of the currently detected road section of 1 kilometer is 93 through the calculation of an intelligent pavement disease identification and detection system.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to be limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention should be determined by the appended claims.
Claims (10)
1. An intelligent pavement disease identification and detection method is characterized by comprising the following steps:
s1, obtaining the position of a vehicle: calculating to obtain the optimal position information of the vehicle according to the current position and speed information of the vehicle;
s2, detecting a road image: obtaining an interested area of a corresponding road image based on the road image data of the current optimal vehicle position information;
s3, extracting road image features: obtaining the characteristics of the corresponding road surface disease image based on the interested area of the road image;
s4, outputting a pavement disease monitoring report form: and outputting a road disease detection report by taking the position and speed information of the vehicle and the characteristics of the corresponding road disease image as input.
2. An intelligent pavement damage identification and detection method according to claim 1, wherein in step S1, based on the current vehicle position and speed information, the speed of the vehicle is subjected to smoothing filtering processing, and the error of the vehicle position information is corrected to obtain the optimal vehicle position information.
3. An intelligent pavement damage recognition and detection method according to claim 2, wherein in step S2, based on the road image data of the current vehicle position information, the acceleration sensor is used to acquire the acceleration data of the vehicle in the vertical direction, and the image position is corrected.
4. The intelligent pavement disease identification and detection method according to claim 3, wherein in step S2, an improved Crack-QuickSort model is adopted to extract the region of interest of the road image, wherein the improved Crack-QuickSort model is composed of a target detection module and a target tracking module; wherein:
adding four serial CrackInclusion blocks in an improved CrackQuickSort model feature extraction layer and removing three full convolution layers, wherein each CrackInclusion block comprises three parallel inclusion-v 3 structures and is used for extracting an interested area of a road image; in the target tracking module, detection frames with the same tracking id are recorded, an image registration algorithm of a longitudinal crack detection frame based on key corner detection is added, image matching is carried out on the longitudinal cracks of the road, and complete detection information of a single longitudinal crack is obtained, wherein the key corner detection adopts an SIFT corner detection algorithm.
5. The intelligent pavement disease identification and detection method according to claim 4, wherein in step S3, the characteristics of the pavement disease image are extracted by using an improved Crack-Wnet algorithm, specifically: adding a super-feature extraction layer between a U-Net neural network coding layer and a decoding layer, wherein the super-feature extraction layer comprises a Dense-block and a W-block which are connected in parallel, the Dense-block comprises two fully-connected layers of 2048 units connected in series, each node in the W-block comprises three convolution layers, the convolution adopts a cavity convolution with the size of 3 x 3, the convolution core expansion rates of the three convolution layers are 1,2,3 respectively, and the super-feature extraction layer is used for enhancing the depth feature extraction capability of road surface disease images.
6. The utility model provides an intelligent road surface disease discernment detecting system which characterized in that includes:
the vehicle position acquisition module is based on a positioning system on a vehicle, takes the current vehicle position and speed information as input, and takes the optimal vehicle position information as output;
the road image detection module is based on a shooting device on the vehicle, takes the road image data of the current vehicle position information as input, and takes the interested area of the corresponding road image as output;
the road image feature extraction module takes the region of interest of the road image as input and the feature of the corresponding road surface disease image as output;
and the road surface defect report output module is used for outputting a defect detection report, and the defect detection report is output by taking the vehicle position and speed information and the characteristics of the corresponding road surface defect image as input.
7. The system of claim 6, wherein the vehicle position obtaining module is configured to: the vehicle position information is input into a vehicle position acquisition module, the speed of the vehicle is subjected to smooth filtering processing, the error of the position information is corrected, and the optimal vehicle position information is obtained.
8. The system of claim 7, wherein the road image detection module is configured to: road image data of current vehicle position information are input to a road image detection module, acceleration data in the vertical direction of the vehicle are obtained by an acceleration sensor, and the position of the image is corrected.
9. The intelligent pavement disease identification and detection system according to claim 8, wherein the road image detection module comprises an improved Crack-quick sort model composed of a target detection module and a target tracking module; adding four serial CrackInception blocks in an improved CrackQuickSort model feature extraction layer and removing three full convolution layers, wherein each CrackInception block comprises three parallel Inception-v3 structures and is used for extracting an interested area of a road image;
in the target tracking module, detection frames with the same tracking id are recorded, an image registration algorithm of a longitudinal crack detection frame based on key corner detection is added, image matching is carried out on the longitudinal cracks of the road, and complete detection information of a single longitudinal crack is obtained, wherein the key corner detection adopts an SIFT corner detection algorithm.
10. The intelligent pavement disease identification and detection system according to claim 9, wherein the road image feature extraction module adds a super feature extraction layer between the U-Net neural network coding layer and the decoding layer by using an improved Crack-wnt algorithm, the super feature extraction layer is a sense-block and a W-block which are connected in parallel, the sense-block is composed of two fully-connected layers of 2048 units connected in series, each node in the W-block is composed of three convolutional layers, the convolution adopts a hole convolution with a size of 3 x 3, and the convolution kernel expansion rates of the three convolutional layers are 1,2,3 respectively, so as to enhance the depth feature extraction capability of the pavement disease image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211165084.5A CN115457277A (en) | 2022-09-23 | 2022-09-23 | Intelligent pavement disease identification and detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211165084.5A CN115457277A (en) | 2022-09-23 | 2022-09-23 | Intelligent pavement disease identification and detection method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115457277A true CN115457277A (en) | 2022-12-09 |
Family
ID=84306195
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211165084.5A Pending CN115457277A (en) | 2022-09-23 | 2022-09-23 | Intelligent pavement disease identification and detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115457277A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116448016A (en) * | 2023-04-26 | 2023-07-18 | 成都智达万应科技有限公司 | Intelligent rapid detection system and detection vehicle with same |
| CN116448773A (en) * | 2023-06-19 | 2023-07-18 | 河北工业大学 | Pavement disease detection method and system with image-vibration characteristics fused |
| CN117522175A (en) * | 2024-01-08 | 2024-02-06 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
-
2022
- 2022-09-23 CN CN202211165084.5A patent/CN115457277A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116448016A (en) * | 2023-04-26 | 2023-07-18 | 成都智达万应科技有限公司 | Intelligent rapid detection system and detection vehicle with same |
| CN116448773A (en) * | 2023-06-19 | 2023-07-18 | 河北工业大学 | Pavement disease detection method and system with image-vibration characteristics fused |
| CN116448773B (en) * | 2023-06-19 | 2023-08-18 | 河北工业大学 | Pavement disease detection method and system with image-vibration characteristics fused |
| CN117522175A (en) * | 2024-01-08 | 2024-02-06 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
| CN117522175B (en) * | 2024-01-08 | 2024-04-02 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115457277A (en) | Intelligent pavement disease identification and detection method and system | |
| CN109325418A (en) | Based on pedestrian recognition method under the road traffic environment for improving YOLOv3 | |
| CN107657226B (en) | People number estimation method based on deep learning | |
| CN109190752A (en) | The image, semantic dividing method of global characteristics and local feature based on deep learning | |
| CN109949340A (en) | Target scale adaptive tracking method based on OpenCV | |
| CN112200143A (en) | Road disease detection method based on candidate area network and machine vision | |
| CN107506765B (en) | License plate inclination correction method based on neural network | |
| CN106599836A (en) | Multi-face tracking method and tracking system | |
| CN111046856B (en) | Parallel pose tracking and map creating method based on dynamic and static feature extraction | |
| CN102169581A (en) | Feature vector-based fast and high-precision robustness matching method | |
| CN109708658B (en) | Visual odometer method based on convolutional neural network | |
| CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
| CN113313031B (en) | Deep learning-based lane line detection and vehicle transverse positioning method | |
| CN113822278A (en) | License plate recognition method for unlimited scene | |
| CN114596500A (en) | Remote sensing image semantic segmentation method based on channel-space attention and DeeplabV3plus | |
| CN112329559A (en) | Method for detecting homestead target based on deep convolutional neural network | |
| CN105404868A (en) | Interaction platform based method for rapidly detecting text in complex background | |
| CN110717900A (en) | Pantograph abrasion detection method based on improved Canny edge detection algorithm | |
| CN113240623A (en) | Pavement disease detection method and device | |
| CN113256683B (en) | Target tracking method and related equipment | |
| CN116704490B (en) | License plate recognition method, license plate recognition device and computer equipment | |
| CN105825215A (en) | Instrument positioning method based on local neighbor embedded kernel function and carrier of method | |
| Han et al. | Accurate and robust vanishing point detection method in unstructured road scenes | |
| CN108763261B (en) | Graph retrieval method | |
| CN116363655A (en) | Financial bill identification method and system |
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 |