CN117095316B - Road surface inspection method, device, equipment and readable storage medium - Google Patents
Road surface inspection method, device, equipment and readable storage medium Download PDFInfo
- Publication number
- CN117095316B CN117095316B CN202311347115.3A CN202311347115A CN117095316B CN 117095316 B CN117095316 B CN 117095316B CN 202311347115 A CN202311347115 A CN 202311347115A CN 117095316 B CN117095316 B CN 117095316B
- Authority
- CN
- China
- Prior art keywords
- image
- defect
- suspected
- road surface
- result
- 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
- 238000007689 inspection Methods 0.000 title claims abstract description 92
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000007547 defect Effects 0.000 claims abstract description 206
- 238000012216 screening Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims description 108
- 230000036541 health Effects 0.000 claims description 26
- 230000002950 deficient Effects 0.000 claims description 22
- 230000000007 visual effect Effects 0.000 claims description 10
- 239000006185 dispersion Substances 0.000 claims description 8
- 238000012552 review Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 abstract description 4
- 238000013102 re-test Methods 0.000 abstract description 3
- 201000010099 disease Diseases 0.000 description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 206010039203 Road traffic accident Diseases 0.000 description 2
- 230000004888 barrier function Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- 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
- G06V10/70—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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The application discloses a road surface inspection method, a device, equipment and a readable storage medium, and relates to the technical field of inspection. In the embodiment of the application, the cloud end can control the unmanned aerial vehicle to move on a road surface according to a preset routing inspection route; acquiring image data of the pavement through the unmanned aerial vehicle; extracting an image frame from the image data, and performing preliminary screening on defects of the image frame; if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result; and marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects. Compare in traditional manual inspection, this application replaces the manual work to accomplish the task of inspecting on road surface through unmanned aerial vehicle, and unmanned aerial vehicle can accomplish the acquisition of ground image data fast to according to image data screening and retest defect, improve the efficiency of inspecting that the road surface inspected greatly, in order to discover defect timely processing.
Description
Technical Field
The application relates to the technical field of inspection, in particular to a pavement inspection method, a pavement inspection device, pavement inspection equipment and a readable storage medium.
Background
Road surface inspection surrounds the discovery, repair, evaluation and decision of road surface diseases, and the discovery of the road surface diseases is the first and most important link in the whole continuous maintenance flow. If the road surface diseases cannot be found timely and accurately, traffic accidents can be caused due to the road surface diseases, casualties, vehicle damage and the like can be caused seriously, and meanwhile, multiple funds and time are needed to be invested for repairing in the later period. The timely discovery of pavement diseases is mainly dependent on manual inspection at present, namely, pavement diseases are discovered manually through naked eyes. It can be understood that the road mileage that needs to patrol is generally longer, and the completion period of manual patrol can be very long at every turn, and the staff is easy tired and has the probability that the staff appears the traffic accident, so patrol inefficiency and with high costs, be difficult to in time discover the road surface disease.
Disclosure of Invention
The main purpose of the application is to provide a road surface inspection method, a device, equipment and a readable storage medium, and aims to solve the technical problems of long completion period of manual inspection, low inspection efficiency and high cost.
In order to achieve the above purpose, the present application provides a road surface inspection method, which includes the following steps:
Controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
acquiring image data of the pavement through the unmanned aerial vehicle;
extracting an image frame from the image data, and performing preliminary screening on defects of the image frame;
if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result;
and marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects.
Optionally, the step of extracting an image frame from the image data includes:
calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
and extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in the corresponding time period of any two adjacent image frames is smaller than or equal to the road surface coverage distance.
Optionally, the step of performing defect preliminary screening on the image frame includes:
determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color of the area is larger than a preset difference threshold value;
Performing non-defect object recognition on an interest target area in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the interest target area is a non-defect object;
determining a first credibility of the first identification result based on each first probability value, and determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest is a non-defective region if the first credibility is greater than a preset threshold value, and is a suspected defective region if the first credibility is less than or equal to the preset threshold value;
if the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
and if the suspected defect area does not exist in the image frame, the preliminary defect screening result of the image frame is that the image frame corresponds to the ground and is first-level health.
Optionally, the step of performing defect review on the suspected image includes:
performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
Determining a second confidence level of the second recognition result based on each second probability value;
if the second credibility is larger than a preset threshold, taking the defect with the largest second probability value in the second identification result as a rechecking result of the suspected image;
and if the credibility is smaller than or equal to the preset threshold value, the rechecking result shows that the road surface corresponding to the suspected image is of secondary health.
Optionally, the first recognition model is trained through a first training sample set, and before the step of recognizing the target region of interest in the image frame based on the preset first recognition model, the method includes:
for any one first training sample in the first training sample set, the label of the first training sample is a road sign, a vehicle or a roadblock;
inputting the first training sample into the first recognition model to obtain a first training recognition result;
updating model parameters of the first recognition model based on differences between the first training recognition result and the labels of the first training samples.
Optionally, the preset second recognition model is trained through a second training sample set, and before the step of performing defect recognition on the suspected defect area in the suspected image based on the preset second recognition model, the method includes:
For any one of the second training samples in the second training sample set, the label of the second training sample is a transverse crack, a longitudinal crack, a crazing or a pit;
inputting the second training sample into the second recognition model to obtain a second training recognition result;
updating model parameters of the second recognition model based on differences between the second training recognition result and the labels of the second training samples.
Optionally, the step of marking the defect in the recheck result on the preset routing includes:
determining a relative position of the suspected image on a time axis of the image data;
and marking defects or secondary health on the preset routing inspection route according to the relative positions.
In addition, in order to realize above-mentioned purpose, this application still provides a road surface inspection device, the road surface inspection device includes:
the control module is used for controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
the acquisition module is used for acquiring the image data of the pavement through the unmanned aerial vehicle;
the extraction module is used for extracting image frames from the image data and carrying out preliminary defect screening on the image frames;
The output module is used for carrying out defect rechecking on the suspected image if the result of the preliminary screening of the defects is the suspected image, and outputting a rechecking result;
and the marking module is used for marking the defects in the rechecking result on the preset inspection route and associating the image frames corresponding to the defects.
In addition, in order to realize above-mentioned purpose, this application still provides a road surface inspection equipment, road surface inspection equipment includes: the road surface inspection system comprises a memory, a processor and a road surface inspection program which is stored in the memory and can run on the processor, wherein the road surface inspection program realizes the steps of the road surface inspection method when being executed by the processor.
In addition, in order to achieve the above object, the present application further provides a readable storage medium, on which a road surface inspection program is stored, which when executed by a processor, implements the steps of the road surface inspection method as described above.
The embodiment of the application provides a road surface inspection method, a device, equipment and a readable storage medium. In the embodiment of the application, the cloud end can control the unmanned aerial vehicle to move on a road surface according to a preset routing inspection route; acquiring image data of the pavement through the unmanned aerial vehicle; extracting an image frame from the image data, and performing preliminary screening on defects of the image frame; if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result; and marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects. Compare in traditional manual inspection, this application replaces the manual work to accomplish the task of inspecting on road surface through unmanned aerial vehicle, and unmanned aerial vehicle can accomplish the acquisition of ground image data fast to according to image data screening and retest defect, improve the efficiency of inspecting that the road surface inspected greatly, in order to discover defect timely processing.
Drawings
FIG. 1 is a schematic diagram of a device architecture of a hardware operating environment according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of a first embodiment of the road inspection method of the present application;
FIG. 3 is a schematic flow chart of a second embodiment of the road inspection method of the present application;
FIG. 4 is a schematic flow chart of a third embodiment of the road inspection method of the present application;
fig. 5 is a schematic structural diagram of the road inspection device.
The realization, functional characteristics and advantages of the present application will be further described with reference to the embodiments, referring to the attached drawings.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
As shown in fig. 1, fig. 1 is a schematic device structure diagram of a hardware running environment according to an embodiment of the present application.
The device of the embodiment of the application can be a mowing robot, and also can be electronic terminal devices such as a smart phone, a PC, a tablet personal computer, a portable computer and the like.
As shown in fig. 1, the apparatus may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
Optionally, the device may also include a camera, RF (Radio Frequency) circuitry, sensors, audio circuitry, wiFi modules, and the like. The terminal may also be configured with other sensors such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., which are not described in detail herein. It will be appreciated by those skilled in the art that the device structure shown in fig. 1 is not limiting of the device and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
It will be appreciated by those skilled in the art that the device structure shown in fig. 1 is not limiting of the device and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
Further, as shown in fig. 1, an operating system, a network communication module, a user interface module, and a road surface inspection program may be included in the memory 1005 as one type of computer-readable storage medium.
In the device shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server, and performing data communication with the background server; the user interface 1003 is mainly used for connecting a user terminal (user terminal) and performing data communication with the user terminal; and the processor 1001 may be configured to call a road inspection program stored in the memory 1005 and perform the following operations:
Controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
acquiring image data of the pavement through the unmanned aerial vehicle;
extracting an image frame from the image data, and performing preliminary screening on defects of the image frame;
if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result;
and marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
the step of extracting an image frame from the image data includes:
calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
and extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in the corresponding time period of any two adjacent image frames is smaller than or equal to the road surface coverage distance.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
The step of performing the preliminary defect screening on the image frame includes:
determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color of the area is larger than a preset difference threshold value;
performing non-defect object recognition on an interest target area in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the interest target area is a non-defect object;
determining a first credibility of the first identification result based on each first probability value, and determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest is a non-defective region if the first credibility is greater than a preset threshold value, and is a suspected defective region if the first credibility is less than or equal to the preset threshold value;
if the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
and if the suspected defect area does not exist in the image frame, the preliminary defect screening result of the image frame is that the image frame corresponds to the ground and is first-level health.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
the step of performing defect review on the suspected image comprises the following steps:
performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
determining a second confidence level of the second recognition result based on each second probability value;
if the second credibility is larger than a preset threshold, taking the defect with the largest second probability value in the second identification result as a rechecking result of the suspected image;
and if the credibility is smaller than or equal to the preset threshold value, the rechecking result shows that the road surface corresponding to the suspected image is of secondary health.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
the first recognition model is trained through a first training sample set, and before the step of recognizing the target region of interest in the image frame based on the preset first recognition model, the method comprises the following steps:
For any one first training sample in the first training sample set, the label of the first training sample is a road sign, a vehicle or a roadblock;
inputting the first training sample into the first recognition model to obtain a first training recognition result;
updating model parameters of the first recognition model based on differences between the first training recognition result and the labels of the first training samples.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
the preset second recognition model is trained through a second training sample set, and before the step of performing defect recognition on the suspected defect area in the suspected image based on the preset second recognition model, the method comprises the following steps:
for any one of the second training samples in the second training sample set, the label of the second training sample is a transverse crack, a longitudinal crack, a crazing or a pit;
inputting the second training sample into the second recognition model to obtain a second training recognition result;
updating model parameters of the second recognition model based on differences between the second training recognition result and the labels of the second training samples.
In a possible implementation, the processor 1001 may call the road inspection program stored in the memory 1005, and further perform the following operations:
the step of marking the defects in the re-inspection result on the preset inspection route comprises the following steps:
determining a relative position of the suspected image on a time axis of the image data;
and marking defects or secondary health on the preset routing inspection route according to the relative positions.
Referring to fig. 2, a first embodiment of a road inspection method of the present application includes:
step S10, controlling the unmanned aerial vehicle to move on a road surface according to a preset routing inspection route;
it should be noted that, in this embodiment, the unmanned aerial vehicle replaces the manual inspection task of completing the road surface, and compared with the manual inspection, the unmanned aerial vehicle can rapidly complete the acquisition of the ground image data, and input the image data into the specially trained recognition model, and the recognition model automatically recognizes the defects existing on the road surface, thereby greatly improving the inspection efficiency of the road surface inspection, and timely finding and timely processing the defects.
The technical staff can set up the route of patrolling and examining for unmanned aerial vehicle according to the road surface demand according to the example, and the aforesaid is preset to patrol and examine the route promptly, and the accessible is preset to patrol and examine the route control unmanned aerial vehicle and remove, also can be with predetermineeing to patrol and examine the route to leading-in unmanned aerial vehicle, by unmanned aerial vehicle according to predetermineeing to patrol and examine the route and remove on the road surface, in addition, also can be by the manual work according to predetermineeing to patrol and examine the route control unmanned aerial vehicle and remove.
Step S20, obtaining image data of the pavement through the unmanned aerial vehicle;
it should be noted that, be provided with visual sensor on the unmanned aerial vehicle, visual sensor's visual field direction can be perpendicular downwards, visual sensor can gather the image data on road surface. The unmanned aerial vehicle can upload the image data that gathers to the high in the clouds. It will be appreciated that the task of identifying defects is typically undertaken by the cloud, as it typically requires a large computational overhead to run the identification model. The cloud will receive image data, such as video, uploaded by the drone.
Step S30, extracting an image frame from the image data, and performing preliminary defect screening on the image frame;
illustratively, an image frame is extracted from the image data. It will be appreciated that if each frame of image in the image data is identified, a significant computational overhead is incurred, and the similarity between any two frames of images in the image data is relatively high, so that it is not necessary to identify each frame of image in the image data. Therefore, the image frames can be extracted from the image data according to a certain interval, and the specific interval can be set by technicians, but the interval is not easy to be too large, so that the omission of the existing ground area is avoided. After the image frames are extracted, the defects of the image frames can be primarily screened by pre-training the identification model. The primary screening mainly identifies some regular non-defect features, such as road signs (e.g. solid lines, broken lines, double solid lines, steering indicating arrows, etc.), vehicles, road blocks, etc., and the remaining features after the non-defect features are screened are suspected defect features. And further identifying the suspected defect features by a rechecking process.
In a possible embodiment, the step of extracting an image frame from the image data includes:
step S311, calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
step S312, extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in the corresponding period of any two adjacent image frames is less than or equal to the road surface coverage distance.
The road surface coverage distance of the image frame is calculated based on the flying height of the unmanned aerial vehicle and the viewing angle span of the unmanned aerial vehicle camera, for example, assuming that the flying height is h and the viewing angle span is a, the road surface coverage distance=2h·tan (a/2). And extracting image frames from the image data according to the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in a period corresponding to any two adjacent image frames is smaller than or equal to the road surface coverage distance. For example, an image frame in which the unmanned aerial vehicle moves after the last image frame (the last extracted image frame) and reaches the road surface coverage distance is extracted, and two adjacent image frames a and b are taken as an example to illustrate that the moving distance is equal to the road surface coverage distance, and the image frame a is the last image frame, and the image frame in which the unmanned aerial vehicle moves from the position corresponding to the image frame a and reaches the road surface coverage distance is the image frame b.
It can be understood that, by the above extraction method, the interval distance between the corresponding positions of two arbitrarily extracted adjacent image frames on the road surface can be the road surface coverage distance of one image frame. On one hand, the connection between two extracted adjacent image frames can be ensured to be tight, and the pavement recognition is avoided. On the other hand, the method also avoids the problem that each image frame in the image data is identified, so that the calculation force is wasted.
In a possible implementation manner, the step of performing the preliminary defect screening on the image frame includes:
step S321, determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color is greater than a preset difference threshold value;
step S322, performing non-defect object recognition on a target area of interest in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the target area of interest is each non-defect object;
step S323, determining a first confidence level of the first recognition result based on each first probability value, and determining a type of the target region of interest according to the first confidence level, wherein the type of the target region of interest is a non-defective region if the first confidence level is greater than a preset threshold value, and is a suspected defective region if the first confidence level is less than or equal to the preset threshold value;
Step S324, if the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
step S325, if the suspected defect area does not exist in the image frame, the result of the preliminary screening of the defects of the image frame is that the ground corresponding to the image frame is first-level health.
For any one image frame, an interest target area is determined from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and the background color is larger than a preset difference threshold value. The background color is usually a normal road surface, for example, gray and black, and may be marked in an image by a technician at the beginning. The determination may also be performed in real time, for example, by determining from a color histogram of the image frame, and using the color with the largest ratio in the color histogram as the background color of the image frame. And taking a region with the background color difference degree larger than a preset difference threshold value in the image frame as the target region of interest.
And carrying out non-defect object recognition on the target region of interest in the image frame through a preset first recognition model to obtain a recognition result, wherein the types of the non-defect objects can comprise non-defect regions such as road signs, vehicles or roadblocks. The first recognition model is preset, and training can be performed in advance to enable the first recognition model to have the capability of recognizing objects such as road signs, vehicles or roadblocks. For any target region of interest, a first recognition model is preset, and the target region of interest can be output as a first probability value of each non-defect object. And determining the first credibility of the first recognition result through each first probability value, wherein the first credibility is in direct proportion to the dispersion of each first probability value, namely, the closer the first probability values are, the higher the first credibility is, for example, the first probability values of road signs, vehicles and roadblocks are respectively 0.1, 0.1 and 0.8, at the moment, the dispersion of each first probability value is high, and the target region of interest has a higher probability of being a roadblock, and accordingly, the first credibility of the first recognition result is higher. On the contrary, if the first probability values of the road sign, the vehicle and the roadblock are respectively 0.3, 0.3 and 0.4, at this time, the dispersion degree of each first probability value is low (relatively close), the first recognition model cannot effectively distinguish whether the target area of interest is the road sign, the vehicle or the roadblock, so the first reliability of the first recognition result is low, and the first reliability can be the variance of each first probability value. And determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest comprises a non-defect region and a suspected defect region. If the first reliability is greater than the preset threshold, the probability that the target region of interest is a non-defective object is greater, and accordingly, the type of the target region of interest can be determined to be a non-defective region. If the first reliability is smaller than or equal to the preset threshold, the probability that the target region of interest is a non-defect object is smaller, and accordingly, the type of the target region of interest can be judged to be a suspected defect region.
For any one image frame, if the image frame has a suspected defect area, the preliminary screening result of the defects of the image frame is a suspected image, otherwise, if the image frame does not have a suspected defect area, the preliminary screening result of the defects of the image frame is that the ground corresponding to the image frame is first-level health.
Step S40, if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result;
for example, if all the features on the image frame are identified as regular non-defect features during the defect preliminary screening, the result of the defect preliminary screening is a normal image. Otherwise, if the image frame has the feature which is not identified as the non-defect feature, the result of the preliminary screening of the defects of the image frame is that the image frame is a suspected image. In this case, the suspected image may be inspected again by another recognition model that recognizes the defect, i.e., the defect review, which may be focused on features that are not recognized as non-defective features. If the corresponding recognition model recognizes the defect characteristics during the defect recheck, the recheck result is that the suspected image has defects corresponding to the ground. Otherwise, if the defect feature is identified, the rechecking result can be that the suspected image corresponds to the ground health.
In a possible embodiment, the step of performing defect review on the suspected image includes:
step S410, performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
step S420, determining a second confidence level of the second recognition result based on each second probability value;
step S430, if the second reliability is greater than a preset threshold, using the defect with the largest second probability value in the second identification result as a recheck result of the suspected image;
step S440, if the reliability is less than or equal to the preset threshold, the rechecking result is that the road surface corresponding to the suspected image is second-level health.
For example, the suspected image may be input to a preset second recognition model, and the preset second recognition model performs defect recognition on the suspected defect area in the suspected image to obtain a second recognition result, where the types of defects may include transverse cracks, longitudinal cracks, crazes, pits, and the like. The second recognition model is preset to be trained in advance to have the ability to recognize transverse cracks, longitudinal cracks, crazes or pits. For any suspected defect area, a second recognition model is preset, and the target area of interest can be output as a second probability value of each defect object. And determining the second credibility of the second recognition result obtained by the second model through each second probability value, wherein the second credibility is in direct proportion to the dispersion of each second probability value, namely, the closer the second probability values are, the higher the second credibility is, for example, the second probability values of transverse cracks, longitudinal cracks, crazes and pits are respectively 0.1, 0.7, at the moment, the dispersion of each first probability value is high, and the suspected defect area has a larger probability of being a pit, and correspondingly, the second credibility of the second recognition result is higher. In contrast, the transverse crack, the longitudinal crack, the crazing and the pit are respectively 0.3, 0.2 and 0.3, and at this time, the dispersion of the second probability values is low (relatively close), and the second recognition model cannot effectively distinguish whether the suspected defect area is the transverse crack, the longitudinal crack, the crazing or the pit, so at this time, the second reliability of the second recognition result is low, and the second reliability may be the variance of the second probability values. And determining a rechecking result according to the second confidence level. If the second confidence level is greater than the preset threshold value, the probability that the suspected defect area is a defect is greater, and accordingly, the defect with the largest second probability value in the second recognition result is taken as a recheck result of the suspected image, for example, based on the above example, the probability of the pit is 0.7, so the recheck result is the pit. If the second reliability is smaller than or equal to the preset threshold, the probability that the suspected defect area is a defect is smaller, and correspondingly, the rechecking result is that the road surface corresponding to the suspected image is second-level health.
It will be appreciated that, in general, the shape of a defect on a road surface is generally random, so that recognition of a model is difficult, and the accuracy of model recognition is improved by two times of recognition (screening and rechecking), for example, the screening can screen out normally occurring features on the road surface, such as road signs and vehicles, so as to avoid recognition errors of the recognition model aiming at the defect during rechecking.
And S50, marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects.
In a possible implementation manner, the step of marking the defect in the recheck result on the preset routing includes:
step S01, determining the relative position of the suspected image on the time axis of the image data;
and S02, marking defects or secondary health on the preset routing inspection route according to the relative positions.
It will be appreciated that the image data may be video and that, correspondingly, there may be a time axis on which the relative position of the suspect image, i.e. the time at which the suspect image appears in the image data, is present. The unmanned aerial vehicle moves at a constant speed generally, so that the position of the suspected image corresponding to the road surface on the preset inspection route and the position of the suspected image on the time axis are in a direct proportion relation, and the defect or the secondary health can be marked on the preset inspection route according to the relative position so as to represent the position of the defect or the secondary health in the actual road surface. It will be appreciated that since the shape of the defect is random, it is difficult for the recognition model to recognize each defect, so that although the road surface is secondary healthy, there may be a defect that the recognition model has not been learned, or there may be garbage in the road surface, so that the suspected image of secondary health may be manually screened again. And marking image frames corresponding to the position-related defects or the secondary health of the defects or the secondary health on a preset inspection route.
In the embodiment, the cloud end can control the unmanned aerial vehicle to move on a road surface according to a preset routing inspection route; acquiring image data of the pavement through the unmanned aerial vehicle; extracting an image frame from the image data, and performing preliminary screening on defects of the image frame; if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result; and marking the defects in the rechecking result on the preset inspection route, and associating the image frames corresponding to the defects. Compare in traditional manual inspection, this application replaces the manual work to accomplish the task of inspecting on road surface through unmanned aerial vehicle, and unmanned aerial vehicle can accomplish the acquisition of ground image data fast to according to image data screening and retest defect, improve the efficiency of inspecting that the road surface inspected greatly, in order to discover defect timely processing.
Referring to fig. 3, a second embodiment of the present application is proposed based on the first embodiment of the present application, and in this embodiment, the same parts as those of the foregoing embodiment may refer to the foregoing, and this embodiment is not repeated. The first recognition model is trained through a first training sample set, and before the step of recognizing the target region of interest in the image frame based on the preset first recognition model, the method comprises the following steps:
Step A10, for any one first training sample in the first training sample set, the label of the first training sample is a road sign, a vehicle or a roadblock;
step A20, inputting the first training sample into the first recognition model to obtain a first training recognition result;
and step A30, updating model parameters of the first recognition model based on the difference between the first training recognition result and the label of the first training sample.
It will be appreciated that if the first recognition model is used to recognize a non-defective object, the first training samples in the first training sample set for training the first recognition model may be images of the non-defective object, for example, the non-defective object may be a road sign, a vehicle, or a road barrier, etc., and the label of the first training sample may be classified as a road sign, a vehicle, or a road barrier, etc. The technician may also set a different first training sample according to the non-defective object that appears on the actual road surface.
And inputting the first training sample into a first recognition model, and recognizing the first training sample by the first recognition model to obtain a first training recognition result. And training the first recognition model based on the difference between the first training recognition result and the first training sample label so as to update model parameters in the first recognition model. For example, the model loss of the first recognition model is calculated by the difference, and model parameters of the first recognition model are updated based on back propagation of the model loss, so that the first recognition model has the capability of recognizing each non-defective object.
Referring to fig. 4, a third embodiment of the present application is proposed based on the first embodiment and the second embodiment of the present application, and in this embodiment, the same parts as those of the foregoing embodiments may refer to the foregoing content, and this embodiment is not repeated. The preset second recognition model is trained through a second training sample set, and before the step of performing defect recognition on the suspected defect area in the suspected image based on the preset second recognition model, the method comprises the following steps:
step B10, for any one of the second training samples in the second training sample set, the label of the second training sample is a transverse crack, a longitudinal crack, a crazing or a pit;
step B20, inputting the second training sample into the second recognition model to obtain a second training recognition result;
and step B30, updating model parameters of the second recognition model based on the difference between the second training recognition result and the label of the second training sample.
For example, if the second recognition model is used to recognize a defect, the second training samples in the second training sample set for training the second recognition model may be images of the defect, for example, the defect may be a transverse crack, a longitudinal crack, a pit, or the like, and the label of the second training sample may be correspondingly classified into a transverse crack, a longitudinal crack, a pit, or the like. The technician may also set a different second training sample based on the presence of a defect in the actual road surface.
And inputting the second training sample into a second recognition model, and recognizing the second training sample by the second recognition model to obtain a second training recognition result. And training the second recognition model based on the difference between the second training recognition result and the second training sample label so as to update model parameters in the second recognition model. For example, model loss of the second recognition model is calculated by the difference, and model parameters of the second recognition model are updated based on back propagation of the model loss, thereby enabling the first recognition model to have the ability to recognize each non-defective object.
In addition, in order to achieve the above object, referring to fig. 5, the present application further provides a road surface inspection device 100, the road surface inspection device 100 including:
the control module 10 is used for controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
an acquisition module 20, configured to acquire image data of the road surface through the unmanned aerial vehicle;
an extracting module 30, configured to extract an image frame from the image data, and perform a preliminary defect screening on the image frame;
the output module 40 is configured to perform a defect review on the suspected image if the result of the preliminary defect screening is the suspected image, and output a review result;
And the marking module 50 is used for marking the defect in the rechecking result on the preset inspection route and associating the image frame corresponding to the defect.
Optionally, the extraction module 30 is further configured to:
calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
and extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in the corresponding time period of any two adjacent image frames is smaller than or equal to the road surface coverage distance.
Optionally, the extraction module 30 is further configured to:
determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color of the area is larger than a preset difference threshold value;
performing non-defect object recognition on an interest target area in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the interest target area is a non-defect object;
determining a first credibility of the first identification result based on each first probability value, and determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest is a non-defective region if the first credibility is greater than a preset threshold value, and is a suspected defective region if the first credibility is less than or equal to the preset threshold value;
If the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
and if the suspected defect area does not exist in the image frame, the preliminary defect screening result of the image frame is that the image frame corresponds to the ground and is first-level health.
Optionally, the rechecking module 40 is further configured to:
performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
determining a second confidence level of the second recognition result based on each second probability value;
if the second credibility is larger than a preset threshold, taking the defect with the largest second probability value in the second identification result as a rechecking result of the suspected image;
and if the credibility is smaller than or equal to the preset threshold value, the rechecking result shows that the road surface corresponding to the suspected image is of secondary health.
Optionally, the first recognition model is trained through a first training sample set, and the road surface inspection device 100 further includes a first training module 60, where the first training module 60 is configured to:
For any one first training sample in the first training sample set, the label of the first training sample is a road sign, a vehicle or a roadblock;
inputting the first training sample into the first recognition model to obtain a first training recognition result;
updating model parameters of the first recognition model based on differences between the first training recognition result and the labels of the first training samples.
Optionally, the preset second recognition model is trained through a second training sample set, and the road surface inspection device 100 further includes a second training module 70, where the second training module 70 is configured to:
for any one of the second training samples in the second training sample set, the label of the second training sample is a transverse crack, a longitudinal crack, a crazing or a pit;
inputting the second training sample into the second recognition model to obtain a second training recognition result;
updating model parameters of the second recognition model based on differences between the second training recognition result and the labels of the second training samples.
Optionally, the marking module 50 is further configured to:
determining a relative position of the suspected image on a time axis of the image data;
And marking defects or secondary health on the preset routing inspection route according to the relative positions.
The application provides a road surface inspection device adopts the road surface inspection method among the above-mentioned embodiment, aims at solving the completion cycle that artifical inspection is long, the technical problem of inspection inefficiency. Compared with the prior art, the beneficial effects of the road surface inspection device provided by the embodiment of the application are the same as those of the road surface inspection method provided by the embodiment, and other technical features of the road surface inspection device are the same as those disclosed by the method of the embodiment, so that the description is omitted.
In addition, in order to realize above-mentioned purpose, this application still provides a road surface inspection equipment, road surface inspection equipment includes: the road surface inspection system comprises a memory, a processor and a road surface inspection program which is stored in the memory and can run on the processor, wherein the road surface inspection program realizes the steps of the road surface inspection method when being executed by the processor.
The specific implementation manner of the device is basically the same as that of each embodiment of the road surface inspection method, and is not repeated here.
In addition, in order to achieve the above object, the present application further provides a readable storage medium, on which a road surface inspection program is stored, which when executed by a processor, implements the steps of the road surface inspection method as described above.
The specific implementation manner of the readable storage medium is basically the same as that of each embodiment of the road surface inspection method, and is not repeated here.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a readable storage medium (such as ROM/RAM, magnetic disk, optical disk) as described above, including several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method described in the embodiments of the present application.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application, or direct or indirect application in other related technical fields are included in the scope of the claims of the present application.
Claims (6)
1. The road surface inspection method is characterized by comprising the following steps of:
controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
acquiring image data of the pavement through the unmanned aerial vehicle;
extracting an image frame from the image data, and performing preliminary screening on defects of the image frame;
if the result of the preliminary screening of the defects is a suspected image, performing defect rechecking on the suspected image, and outputting a rechecking result;
marking the defects in the re-inspection result on the preset inspection route, and associating the image frames corresponding to the defects;
the step of performing defect preliminary screening on the image frame comprises the following steps:
determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color of the area is larger than a preset difference threshold value;
Performing non-defect object recognition on an interest target area in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the interest target area is a non-defect object;
determining a first credibility of the first identification result based on each first probability value, and determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest is a non-defective region if the first credibility is greater than a preset threshold value, the type of the target region of interest is a suspected defective region if the first credibility is less than or equal to the preset threshold value, and the first credibility is proportional to the dispersion of each first probability value;
if the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
if the suspected defect area does not exist in the image frame, the preliminary defect screening result of the image frame is that the ground corresponding to the image frame is first-level health;
wherein the step of extracting an image frame from the image data comprises:
Calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in a period corresponding to any two adjacent image frames is smaller than or equal to the road surface coverage distance;
the step of performing defect review on the suspected image comprises the following steps:
performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
determining a second confidence level of the second recognition result based on each second probability value;
if the second credibility is larger than a preset threshold, taking the defect with the largest second probability value in the second identification result as a rechecking result of the suspected image;
if the credibility is smaller than or equal to the preset threshold value, the rechecking result shows that the road surface corresponding to the suspected image is of secondary health;
the step of marking the defect in the rechecking result on the preset routing inspection route comprises the following steps:
Determining a relative position of the suspected image on a time axis of the image data;
and marking defects or secondary health on the preset routing inspection route according to the relative positions.
2. The method of claim 1, wherein the first recognition model is trained with a first set of training samples, the method comprising, prior to the step of recognizing the target region of interest in the image frame based on a preset first recognition model:
for any one first training sample in the first training sample set, the label of the first training sample is a road sign, a vehicle or a roadblock;
inputting the first training sample into the first recognition model to obtain a first training recognition result;
updating model parameters of the first recognition model based on differences between the first training recognition result and the labels of the first training samples.
3. The method of claim 1, wherein the predetermined second recognition model is trained with a second training sample set, and wherein prior to the step of defect recognition of the suspected defective area in the suspected image based on the predetermined second recognition model, the method comprises:
For any one of the second training samples in the second training sample set, the label of the second training sample is a transverse crack, a longitudinal crack, a crazing or a pit;
inputting the second training sample into the second recognition model to obtain a second training recognition result;
updating model parameters of the second recognition model based on differences between the second training recognition result and the labels of the second training samples.
4. The utility model provides a road surface inspection device which characterized in that, the road surface inspection device includes:
the control module is used for controlling the unmanned aerial vehicle to move on the road surface according to a preset routing inspection route;
the acquisition module is used for acquiring the image data of the pavement through the unmanned aerial vehicle;
the extraction module is used for extracting image frames from the image data and carrying out preliminary defect screening on the image frames;
the output module is used for carrying out defect rechecking on the suspected image if the result of the preliminary screening of the defects is the suspected image, and outputting a rechecking result;
the marking module is used for marking the defects in the rechecking result on the preset routing inspection route and associating the image frames corresponding to the defects;
The step of performing defect preliminary screening on the image frame comprises the following steps:
determining an interest target area from the image frame, wherein the interest target area is an area in the image frame, and the difference degree between the interest target area and a background color of the area is larger than a preset difference threshold value;
performing non-defect object recognition on an interest target area in the image frame based on a preset first recognition model to obtain a first recognition result, wherein the first recognition result comprises a first probability value that the interest target area is a non-defect object;
determining a first credibility of the first identification result based on each first probability value, and determining the type of the target region of interest according to the first credibility, wherein the type of the target region of interest is a non-defective region if the first credibility is greater than a preset threshold value, the type of the target region of interest is a suspected defective region if the first credibility is less than or equal to the preset threshold value, and the first credibility is proportional to the dispersion of each first probability value;
if the suspected defect area exists in the image frame, the preliminary defect screening result of the image frame is a suspected image;
If the suspected defect area does not exist in the image frame, the preliminary defect screening result of the image frame is that the ground corresponding to the image frame is first-level health;
wherein the step of extracting an image frame from the image data comprises:
calculating the road surface coverage distance of the image frame according to the flying height of the unmanned aerial vehicle and the visual angle span of the unmanned aerial vehicle camera;
extracting the image frames from the image data based on the road surface coverage distance, wherein the moving distance of the unmanned aerial vehicle in a period corresponding to any two adjacent image frames is smaller than or equal to the road surface coverage distance;
the step of performing defect review on the suspected image comprises the following steps:
performing defect recognition on a suspected defect area in the suspected image based on a preset second recognition model to obtain a second recognition result, wherein the second recognition result comprises a second probability value that the suspected defect area is each defect;
determining a second confidence level of the second recognition result based on each second probability value;
if the second credibility is larger than a preset threshold, taking the defect with the largest second probability value in the second identification result as a rechecking result of the suspected image;
If the credibility is smaller than or equal to the preset threshold value, the rechecking result shows that the road surface corresponding to the suspected image is of secondary health;
the step of marking the defect in the rechecking result on the preset routing inspection route comprises the following steps:
determining a relative position of the suspected image on a time axis of the image data;
and marking defects or secondary health on the preset routing inspection route according to the relative positions.
5. A pavement inspection apparatus, comprising: a memory, a processor and a road surface inspection program stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the road surface inspection method of any one of claims 1 to 3.
6. A readable storage medium, wherein a road surface inspection program is stored on the readable storage medium, which when executed by a processor, implements the steps of the road surface inspection method according to any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311347115.3A CN117095316B (en) | 2023-10-18 | 2023-10-18 | Road surface inspection method, device, equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311347115.3A CN117095316B (en) | 2023-10-18 | 2023-10-18 | Road surface inspection method, device, equipment and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117095316A CN117095316A (en) | 2023-11-21 |
| CN117095316B true CN117095316B (en) | 2024-02-09 |
Family
ID=88772048
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311347115.3A Active CN117095316B (en) | 2023-10-18 | 2023-10-18 | Road surface inspection method, device, equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117095316B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126460A (en) * | 2019-12-10 | 2020-05-08 | 福建省高速公路集团有限公司 | Pavement disease automatic inspection method, medium, equipment and device based on artificial intelligence |
| CN112288711A (en) * | 2020-10-28 | 2021-01-29 | 浙江华云清洁能源有限公司 | Unmanned aerial vehicle inspection image defect image identification method, device, equipment and medium |
| CN113888462A (en) * | 2021-08-27 | 2022-01-04 | 中国电力科学研究院有限公司 | Crack identification method, system, readable medium and storage medium |
| CN114445411A (en) * | 2022-04-11 | 2022-05-06 | 广东电网有限责任公司佛山供电局 | Unmanned aerial vehicle line patrol defect identification system and control method |
| CN114529545A (en) * | 2022-04-22 | 2022-05-24 | 天津理工大学 | Unmanned aerial vehicle-based road defect automatic detection method and system |
| CN115994901A (en) * | 2023-02-13 | 2023-04-21 | 北京理工大学前沿技术研究院 | Automatic road disease detection method and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160292518A1 (en) * | 2015-03-30 | 2016-10-06 | D-Vision C.V.S Ltd | Method and apparatus for monitoring changes in road surface condition |
-
2023
- 2023-10-18 CN CN202311347115.3A patent/CN117095316B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126460A (en) * | 2019-12-10 | 2020-05-08 | 福建省高速公路集团有限公司 | Pavement disease automatic inspection method, medium, equipment and device based on artificial intelligence |
| CN112288711A (en) * | 2020-10-28 | 2021-01-29 | 浙江华云清洁能源有限公司 | Unmanned aerial vehicle inspection image defect image identification method, device, equipment and medium |
| CN113888462A (en) * | 2021-08-27 | 2022-01-04 | 中国电力科学研究院有限公司 | Crack identification method, system, readable medium and storage medium |
| CN114445411A (en) * | 2022-04-11 | 2022-05-06 | 广东电网有限责任公司佛山供电局 | Unmanned aerial vehicle line patrol defect identification system and control method |
| CN114529545A (en) * | 2022-04-22 | 2022-05-24 | 天津理工大学 | Unmanned aerial vehicle-based road defect automatic detection method and system |
| CN115994901A (en) * | 2023-02-13 | 2023-04-21 | 北京理工大学前沿技术研究院 | Automatic road disease detection method and system |
Non-Patent Citations (1)
| Title |
|---|
| 基于无人机图像识别技术的输电线路缺陷检测;李宁 等;电子设计工程(10);第105-109页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117095316A (en) | 2023-11-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jiang et al. | Real‐time crack assessment using deep neural networks with wall‐climbing unmanned aerial system | |
| KR102121958B1 (en) | Method, system and computer program for providing defect analysis service of concrete structure | |
| WO2020199538A1 (en) | Bridge key component disease early-warning system and method based on image monitoring data | |
| KR102094341B1 (en) | System for analyzing pot hole data of road pavement using AI and for the same | |
| CN111797890A (en) | Method and system for detecting defects of power transmission line equipment | |
| Akagic et al. | Pothole detection: An efficient vision based method using rgb color space image segmentation | |
| CN102073846B (en) | Method for acquiring traffic information based on aerial images | |
| CN112115927B (en) | Intelligent machine room equipment identification method and system based on deep learning | |
| CN114998852A (en) | Intelligent detection method for road pavement diseases based on deep learning | |
| CN111274926B (en) | Image data screening method, device, computer equipment and storage medium | |
| CN104657706A (en) | Image-based high-speed railway line pole breakage abnormality and connecting structural body abnormality detection method | |
| CN113052295B (en) | Training method of neural network, object detection method, device and equipment | |
| CN112528979B (en) | Transformer substation inspection robot obstacle distinguishing method and system | |
| CN113688817A (en) | Instrument identification method and system for automatic inspection | |
| CN113077416A (en) | Welding spot welding defect detection method and system based on image processing | |
| CN115830399A (en) | Classification model training method, apparatus, device, storage medium, and program product | |
| CN113706496B (en) | Aircraft structure crack detection method based on deep learning model | |
| CN111062437A (en) | Bridge structure disease automatic target detection model based on deep learning | |
| CN117095316B (en) | Road surface inspection method, device, equipment and readable storage medium | |
| CN112508911A (en) | Rail joint touch net suspension support component crack detection system based on inspection robot and detection method thereof | |
| CN112184903A (en) | Method, device, equipment and medium for detecting high-voltage line tree obstacle risk points | |
| CN115995056A (en) | Automatic bridge disease identification method based on deep learning | |
| CN115994901A (en) | Automatic road disease detection method and system | |
| WO2024000372A1 (en) | Defect detection method and apparatus | |
| CN115546191A (en) | Insulator defect detection method and device based on improved RetinaNet |
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 |