CN212084165U - Equipment for detecting cracks of asphalt road pavement based on neural network - Google Patents
Equipment for detecting cracks of asphalt road pavement based on neural network Download PDFInfo
- Publication number
- CN212084165U CN212084165U CN202020430316.5U CN202020430316U CN212084165U CN 212084165 U CN212084165 U CN 212084165U CN 202020430316 U CN202020430316 U CN 202020430316U CN 212084165 U CN212084165 U CN 212084165U
- Authority
- CN
- China
- Prior art keywords
- neural network
- server
- wireless transmission
- transmission module
- internet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The utility model discloses a detect cracked equipment of bituminous paving based on neural network, concretely relates to computer vision field, including aircraft, the cloud platform of making a video recording, wireless transmission module and server, the output of the cloud platform of making a video recording and wireless transmission module's input electric connection, wireless transmission module's output and the input electric signal connection of server, wireless transmission module is used for carrying out real-time transmission to the server with the bituminous paving image information that the cloud platform gathered of making a video recording, and the inside of server is equipped with convolution neural network, attention module. The utility model discloses a degree of depth convolution neural network draws the characteristic, combines together artificial neural network and degree of depth learning technique, has the global training characteristic that structure level, characteristic extraction and classification combine, utilizes the technique of weight sharing, has reduced the parameter in the network, overcomes the big scheduling problem of calculated amount among the neural network, and it is strong to have the generalization ability, characteristics that recognition efficiency is high.
Description
Technical Field
The utility model relates to a computer vision technical field, more specifically say, the utility model specifically is a detect cracked equipment on bituminous paving based on neural network.
Background
The visual system is an important way for human to perceive the world and is the most deeply known part of human research, and the research shows that 80% of external information is transmitted to the brain through the human visual system. In real life, people can easily perceive the world through a vision system. Although the human visual system is very powerful, it is still necessary to develop computer vision that can simulate the human visual system. Currently, computer vision technology is receiving more and more attention from students and achieving abundant results, but is far away from the target of perceiving external information like human vision. Target detection is an important research direction of computer vision, and some technologies have been successfully applied to our real life, such as face recognition, pedestrian detection, and the like. As society gradually enters the intelligent and information age, the division and identification of scene objects will play an increasingly important role in future life.
Conventional target detection is generally divided into three stages: firstly, some candidate regions are selected on a given image, then the regions are subjected to feature extraction, and finally, a trained classifier is used for classification. There are two main problems with the traditional target detection task: one is that the region selection strategy based on the sliding window has no pertinence, the time complexity is high, and the window is redundant; secondly, the manually designed features are not very robust to variations in diversity. The two problems can be well solved by the deep learning target detection algorithm based on the RegionProposal, the RegionProposal finds out the position where the target possibly appears in the image in advance, and the RegionProposal utilizes the information such as texture, edge, color and the like in the image, so that the higher recall rate can be kept under the condition of selecting fewer windows, the time complexity of subsequent operation is greatly reduced, and the quality of the obtained candidate window is higher than that of a sliding window.
Most of cracks of the traditional detection asphalt road surface are inspected on the spot by personnel or are reacted by the masses to acquire information data, so that the detection method has time delay, a large amount of operation cost is huge in the inspection on the spot, and in addition, although the novel satellite remote sensing image recognition is simpler and easier than the mode of the inspection on the spot, the recognition efficiency is too low, and the accuracy is far from the monitoring standard.
Therefore, the device for detecting the cracks of the asphalt road pavement based on the neural network is needed to be provided.
SUMMERY OF THE UTILITY MODEL
In order to overcome the above defects of the prior art, the embodiment of the utility model provides a crack's equipment of detection bituminous paving based on neural network, extract the characteristic through the deep convolution neural network, combine artificial neural network and deep learning technique, have the global training characteristic that structure level, characteristic extraction and classification combine, utilize the technique of weight sharing, reduced the parameter in the network, overcome the big scheduling problem of calculated amount in the neural network, have the characteristics that generalization ability is strong, recognition efficiency is high; additionally, the utility model discloses a filtration of attention mechanism, under the condition of a given characteristic map, along two solitary dimensions in passageway and space in proper order generate attention map, then will notice the picture and take advantage of and carry out the adaptive feature in the input characteristic map and refine, more refined the characteristic that the convolution network extracted, further improved the recognition accuracy of model to solve the problem that proposes in the above-mentioned background art.
In order to achieve the above object, the utility model provides a following technical scheme: the utility model provides an equipment for detecting crack on bituminous paving based on neural network, includes the aircraft, makes a video recording the cloud platform, wireless transmission module and server, the aircraft is used for carrying the cloud platform of making a video recording and carries out high altitude investigation and shoot and gather bituminous paving image information, the output of the cloud platform of making a video recording and wireless transmission module's input electric connection, wireless transmission module's output and the input electric signal connection of server, wireless transmission module is used for carrying out real-time transmission to the server with the bituminous paving image information that the cloud platform gathered of making a video recording, the inside of server is equipped with convolution neural network, attention module, the server is used for saving and carrying out image recognition and generating detection information according to convolution neural network, attention module algorithm to the bituminous paving image information that the cloud platform gathered of making a video recording.
In a preferred embodiment, the maximum flying height range of the aircraft is less than 330 meters, and the control end of the aircraft is electrically connected with a flying controller through signals, and the flying controller can wirelessly control the camera head to record images.
In a preferred embodiment, the wireless transmission module is a wireless transmission module based on communication of the internet of things, an output end of the wireless transmission module is electrically connected with a communication base station, and an input end and an output end of the communication base station are connected with an output end and an input end of the server in a wired or wireless manner.
In a preferred embodiment, the communication base station obtains attribute reporting information from at least one terminal device of the internet of things, where each of the attribute reporting information includes: the identification ID of the terminal equipment of the Internet of things; the communication base station divides the at least one terminal device of the internet of things into at least one group, one terminal device of the internet of things is selected from each group to serve as a sink node, and the rest terminal devices of the internet of things serve as member nodes; the communication base station sends a service reporting instruction to the at least one internet of things terminal device, wherein the service reporting instruction comprises: wireless communication resources, and a correspondence between the sink node and the member node.
The utility model discloses a technological effect and advantage:
1. the utility model extracts the features through the deep convolution neural network, combines the artificial neural network with the deep learning technology, has the global training features of structure level, feature extraction and classification combination, utilizes the technology of weight sharing, reduces the parameters in the network, overcomes the problems of large calculation amount and the like in the neural network, and has the characteristics of strong generalization capability and high recognition efficiency;
2. the utility model discloses a filtration of attention mechanism under the condition of a given characteristic map, generates attention mapping in proper order along two solitary dimensions in passageway and space, then will notice the picture and take advantage of and carry out the adaptive feature in the input characteristic map and refine, has refined the characteristic that the convolution network drawed more, has further improved the recognition accuracy of model.
Drawings
Fig. 1 is a schematic view of the overall structure of the present invention.
Fig. 2 is a schematic diagram of the image information transmission structure of the present invention.
Fig. 3 is a schematic diagram of the system model structure of the present invention.
Fig. 4 is a schematic diagram of the algorithm flow structure of the present invention.
The reference signs are: 1. an aircraft; 2. a camera shooting cloud deck; 3. a wireless transmission module; 4. a server; 5. A convolutional neural network; 6. an attention module.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments in the present invention, all other embodiments obtained by a person skilled in the art without creative work belong to the protection scope of the present invention.
The equipment for detecting the cracks of the asphalt road pavement based on the neural network as shown in the attached drawings 1-4 comprises an aircraft 1, a camera platform 2, a wireless transmission module 3 and a server 4, wherein the aircraft 1 is used for carrying the camera platform 2 to perform high-altitude detection, shooting and collecting the image information of the asphalt road pavement, the output end of the camera platform 2 is electrically connected with the input end of the wireless transmission module 3, the output end of the wireless transmission module 3 is in electric signal connection with the input end of the server 4, the wireless transmission module 3 is used for transmitting the image information of the asphalt road pavement, collected by the camera platform 2, to the server 4 in real time, the server 4 is internally provided with a convolutional neural network 5 and an attention module 6, the server 4 is used for storing the image information of the asphalt road pavement collected by the camera head 2, identifying the image according to the algorithm of the convolutional neural network 5 and the attention module 6 and generating detection information.
The implementation mode is specifically as follows: the features are extracted through a deep convolutional neural network, an artificial neural network and a deep learning technology are combined, the global training features of combination of structural hierarchy, feature extraction and classification are realized, the weight sharing technology is utilized, the parameters in the network are reduced, the problems of large calculation amount and the like in the neural network are solved, and the method has the characteristics of strong generalization capability and high recognition efficiency; additionally, the utility model discloses a filtration of attention mechanism, under the condition of a given characteristic map, along two solitary dimensions in passageway and space in proper order generate attention map, then will notice the picture and take advantage of and carry out the adaptive feature in the input characteristic map and refine, refine the characteristic that the convolution network extracted more, further improved the recognition accuracy of model.
The maximum flight height range of the aircraft 1 is less than 330 meters, the control end of the aircraft 1 is connected with a flight controller through an electric signal, and the flight controller can wirelessly control the camera head 2 to shoot and record images, so that the accuracy of shooting and recording the images is guaranteed.
Wherein, wireless transmission module 3 is based on the wireless transmission module of thing networking communication, and wireless transmission module 3's output electric signal connection has the communication base station, and the input and the output of communication base station adopt thing to ally oneself with communication technology through wired or wireless mode and server 4's output and input, and its cost is cheaper, is applicable to the regional image acquisition of large tracts of land.
The communication base station acquires attribute reporting information from at least one terminal device of the internet of things, wherein each attribute reporting information comprises: identification ID of the terminal equipment of the Internet of things; the communication base station divides at least one terminal device of the Internet of things into at least one group, one terminal device of the Internet of things is selected from each group as a sink node, and the rest terminal devices of the Internet of things are used as member nodes; the communication base station sends a service reporting instruction to at least one terminal device of the Internet of things, wherein the service reporting instruction comprises the following steps: and the wireless communication resources, the corresponding relation between the aggregation nodes and the member nodes realize an internet of things communication architecture.
The server 4 receives the image information of the asphalt road surface transmitted by the camera holder 2, performs tensor conversion and executes image data feature extraction, and the convolutional neural network 5 and the attention module 6 act on the whole operation process of the server 4.
The utility model discloses the theory of operation:
firstly, acquiring an original image, converting the image into a tensor form required by a convolutional neural network, extracting image characteristics by the convolutional neural network, learning image data characteristics layer by layer in a mode of arranging a plurality of serial convolutional layers and pooling layers at intervals, scanning the whole image by using a convolutional kernel smaller than the size of the image in a convolutional operation mode, and calculating the sum of the convolutional kernel and the local position weight of the image; each convolution corresponds to a feature map and is then input to the pooling layer for spatial subsampling, so that the convolutional neural network has a certain distortion resistance. The top layer of the network pulls all the obtained feature mappings into one-dimensional vectors again, network parameters are adjusted by combining a multi-classification regression classifier with a reverse propagation error signal, then the FRCNN convolutional neural network is used for detecting the features of the pavement cracks from the original image, and the optimized crack features are obtained through an attention module; and finally, the optimized features are subjected to a classifier to obtain a correct detection result.
The points to be finally explained are: first, in the description of the present application, it should be noted that, unless otherwise specified and limited, the terms "mounted," "connected," and "connected" should be understood broadly, and may be a mechanical connection or an electrical connection, or a communication between two elements, and may be a direct connection, and "upper," "lower," "left," and "right" are only used to indicate a relative positional relationship, and when the absolute position of the object to be described is changed, the relative positional relationship may be changed;
secondly, the method comprises the following steps: in the drawings of the disclosed embodiments of the present invention, only the structures related to the disclosed embodiments are referred to, and other structures can refer to the common design, and under the condition of no conflict, the same embodiment and different embodiments of the present invention can be combined with each other;
and finally: the above description is only for the preferred embodiment of the present invention and should not be taken as limiting the invention, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (4)
1. The utility model provides an equipment of crack of detection bituminous paving road surface which characterized in that based on neural network: including aircraft (1), the cloud platform (2) of making a video recording, wireless transmission module (3) and server (4), aircraft (1) is used for carrying the cloud platform (2) of making a video recording and carries out high altitude investigation and shoot and gather bituminous paving image information, the output of the cloud platform (2) of making a video recording and the input electric connection of wireless transmission module (3), the output of wireless transmission module (3) and the input signal connection of server (4), wireless transmission module (3) are used for carrying out real-time transmission to server (4) with the bituminous paving image information that cloud platform (2) of making a video recording gathered, the inside of server (4) is equipped with convolution neural network (5), attention module (6), server (4) are used for storing and according to convolution neural network (5) bituminous paving image information that cloud platform (2) of making a video recording gathered, The attention module (6) algorithm performs image recognition and generates detection information.
2. The apparatus for detecting cracks of asphalt road pavement based on the neural network as claimed in claim 1, wherein: the maximum flight height range of the aircraft (1) is less than 330 meters, the control end of the aircraft (1) is connected with a flight controller through an electric signal, and the flight controller can wirelessly control the camera head (2) to shoot and record images.
3. The apparatus for detecting cracks of asphalt road pavement based on the neural network as claimed in claim 1, wherein: the wireless transmission module (3) is a wireless transmission module based on internet of things communication, the output end of the wireless transmission module (3) is connected with a communication base station in an electric signal mode, and the input end and the output end of the communication base station are connected with the output end and the input end of the server (4) in a wired or wireless mode.
4. The apparatus for detecting cracks of asphalt road pavement based on the neural network as claimed in claim 3, wherein: the communication base station acquires attribute reporting information from at least one terminal device of the internet of things, wherein each attribute reporting information comprises: the identification ID of the terminal equipment of the Internet of things; the communication base station divides the at least one terminal device of the internet of things into at least one group, one terminal device of the internet of things is selected from each group to serve as a sink node, and the rest terminal devices of the internet of things serve as member nodes; the communication base station sends a service reporting instruction to the at least one internet of things terminal device, wherein the service reporting instruction comprises: wireless communication resources, and a correspondence between the sink node and the member node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202020430316.5U CN212084165U (en) | 2020-03-30 | 2020-03-30 | Equipment for detecting cracks of asphalt road pavement based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202020430316.5U CN212084165U (en) | 2020-03-30 | 2020-03-30 | Equipment for detecting cracks of asphalt road pavement based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN212084165U true CN212084165U (en) | 2020-12-04 |
Family
ID=73565040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202020430316.5U Active CN212084165U (en) | 2020-03-30 | 2020-03-30 | Equipment for detecting cracks of asphalt road pavement based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN212084165U (en) |
-
2020
- 2020-03-30 CN CN202020430316.5U patent/CN212084165U/en active Active
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021088300A1 (en) | Rgb-d multi-mode fusion personnel detection method based on asymmetric double-stream network | |
| CN106356757B (en) | A kind of power circuit unmanned plane method for inspecting based on human-eye visual characteristic | |
| CN108109385B (en) | System and method for identifying and judging dangerous behaviors of power transmission line anti-external damage vehicle | |
| CN109118479A (en) | Defects of insulator identification positioning device and method based on capsule network | |
| CN106682592B (en) | Image automatic identification system and method based on neural network method | |
| CN110070530A (en) | A kind of powerline ice-covering detection method based on deep neural network | |
| CN102053563A (en) | Flight training data acquisition and quality evaluation system of analog machine | |
| CN108921039A (en) | The forest fire detection method of depth convolution model based on more size convolution kernels | |
| CN110991466A (en) | Highway road surface condition detecting system based on novel vision sensing equipment | |
| CN108648169A (en) | The method and device of high voltage power transmission tower defects of insulator automatic identification | |
| CN109631776B (en) | Automatic measurement method for icing thickness of high-voltage transmission line conductor | |
| CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
| CN110245588A (en) | A kind of fine granularity estimation method of human posture based on radio frequency signal | |
| CN109086803B (en) | Deep learning and personalized factor-based haze visibility detection system and method | |
| CN112184628B (en) | Infrared duplex wave image and cloud early warning system and method for flood prevention and danger detection of dike | |
| CN101833673A (en) | Electric power switchgear switch state image recognition system | |
| CN114089786A (en) | Autonomous inspection system based on unmanned aerial vehicle vision and along mountain highway | |
| CN113947555A (en) | Infrared and visible light fused visual system and method based on deep neural network | |
| CN114240868A (en) | Unmanned aerial vehicle-based inspection analysis system and method | |
| CN116719339A (en) | Unmanned aerial vehicle-based power line inspection control method and system | |
| CN109359545B (en) | Cooperative monitoring method and device under complex low-altitude environment | |
| CN108563986B (en) | Method and system for judging posture of telegraph pole in jolt area based on long-distance shooting image | |
| CN110033045A (en) | A kind of method and apparatus of trained identification image atomization | |
| CN212084165U (en) | Equipment for detecting cracks of asphalt road pavement based on neural network | |
| CN114592411B (en) | Carrier parasitic type intelligent inspection method for highway damage |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant |