CN117788950A - Pavement disease detection method based on improved YOLOv8 model - Google Patents
Pavement disease detection method based on improved YOLOv8 model Download PDFInfo
- Publication number
- CN117788950A CN117788950A CN202410021694.0A CN202410021694A CN117788950A CN 117788950 A CN117788950 A CN 117788950A CN 202410021694 A CN202410021694 A CN 202410021694A CN 117788950 A CN117788950 A CN 117788950A
- Authority
- CN
- China
- Prior art keywords
- module
- yolov8
- model
- dcnv2
- improved
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 82
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 82
- 238000001514 detection method Methods 0.000 title claims abstract description 67
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 238000012795 verification Methods 0.000 claims abstract description 9
- 230000007246 mechanism Effects 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 238000010200 validation analysis Methods 0.000 claims description 2
- 230000001788 irregular Effects 0.000 abstract description 6
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 8
- 230000007547 defect Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000012423 maintenance Methods 0.000 description 4
- 230000003321 amplification Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000003199 nucleic acid amplification method Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 235000002566 Capsicum Nutrition 0.000 description 1
- 239000006002 Pepper Substances 0.000 description 1
- 235000016761 Piper aduncum Nutrition 0.000 description 1
- 235000017804 Piper guineense Nutrition 0.000 description 1
- 244000203593 Piper nigrum Species 0.000 description 1
- 235000008184 Piper nigrum Nutrition 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 239000010426 asphalt Substances 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000012633 leachable Substances 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a pavement disease detection method based on an improved YOLOv8 model, which comprises the steps of firstly obtaining a pavement disease image data set and dividing the pavement disease image data set into a training set, a verification set and a test set; then constructing an improved YOLOv8 model that introduces a deformable convolution and receptive field attention mechanism; then training the improved YOLOv8 model based on the training set, the verification set and the test set to obtain a trained improved YOLOv8 model; and finally, inputting the pavement disease image to be detected into a trained improved YOLOv8 model for recognition, and obtaining a detection result. The invention introduces deformable convolution, can adaptively adjust and adapt to irregular disease shapes, and better captures local characteristics; meanwhile, a receptive field attention mechanism is added, so that the receptive field of the neural network can be enhanced, the detection of road surface diseases with different dimensions is adapted, and the interference of complex backgrounds is overcome.
Description
Technical Field
The invention relates to the technical field of information perception and identification, in particular to a pavement disease detection method based on an improved YOLOv8 model.
Background
The highway construction in China develops rapidly, and reaches the end of 2022, and the total mileage of highway traffic reaches 535.48 ten thousand kilometers. Along with the continuous perfection of the road mileage construction in China, the requirements for road maintenance service are also increasing. The highway in China is mostly an asphalt pavement, and due to the defects of large gap, poor temperature stability, poor aging resistance and the like, the pavement is extremely easy to cause diseases such as cracks, looseness and the like, and the concrete pavement can generate pavement diseases due to overlarge traffic bearing capacity and severe environments such as road ponding and the like. The road surface diseases affect the attractiveness of the road, reduce the comfort of driving personnel, reduce the bearing capacity and the transportation capacity of the road and the service life of the road, and even threaten the driving safety of passenger and cargo transportation, cause traffic accidents and cause serious casualties and economic losses. Therefore, the rapid, accurate and effective detection of the pavement diseases is important for guaranteeing the road transportation safety.
The traditional manual detection means directly observe road surface diseases through inspection personnel using eyes or small auxiliary equipment, the potential safety hazard of the method is large, the data collection is incomplete and inaccurate, the detection result is easily subjectively influenced by the inspection personnel, the detection efficiency is low, because the total road mileage is longer, the manual detection labor cost is higher, and the road is closed during the detection to cause traffic jam to influence traffic. Road surface disease detection based on vision is carried out by collecting road images through equipment such as a vehicle recorder, a vehicle-mounted camera and an unmanned aerial vehicle, has low cost, has the characteristics of no damage, high precision and high efficiency, is widely applied in recent years, but in actual work, the rapid and accurate identification of road surface diseases is still a difficult problem. The one-stage target detection network is superior to the two-stage target detection network in detection speed, so that the one-stage target detection network is widely applied to the fields with high real-time requirements such as industrial detection. The YOLO series model has been attracting attention in one-stage network, and the current latest YOLO series model YOLOv8 has been less studied in road surface detection.
Therefore, the technical problem to be solved is how to optimize the basic network structure by means of the YOLOv8 model so as to be suitable for pavement disease detection and improve the detection speed and accuracy.
Disclosure of Invention
Aiming at the defects related to the background technology, the invention provides a pavement disease detection method based on an improved YOLOv8 model, which can effectively solve the problems of irregular disease shape, large difference of disease size, weak disease information, complex pavement background interference and the like, and the problem that the original YOLOv8 detection network cannot fully learn relevant characteristics of the disease, so that the detection accuracy is low.
The invention adopts the following technical scheme for solving the technical problems:
the pavement disease detection method based on the improved YOLOv8 model comprises the following steps:
step 1), obtaining a pavement disease image data set and dividing the pavement disease image data set into a training set, a verification set and a test set;
step 2), constructing an improved YOLOv8 model which introduces a deformable convolution and receptive field attention mechanism;
step 2.1), constructing a C2f_dcnv2 module;
the C2f_dcnv2 module comprises a first Conv module, a second Conv module, a Split module and a first Bottleneck_dcnv2 module;
the first to third Bottleneck_dcnv2 modules have the same structure and comprise a third Conv module, a deformable convolution layer DCNv2, a batch normalization layer and a SiLU activation function layer which are sequentially connected, wherein the input end of the third Conv module is connected with external input, and the input end of the third Conv module and the output end of the SiLU activation function layer are added and then used as output;
the first Conv module, the Split module, the first Bottleneck_dcnv2 module, the second Bottleneck_dcnv2 module and the third Bottleneck_dcnv2 module are sequentially connected, the input end of the first Conv module is used as the input end of the C2f_dcnv2 module, the output end of the Split module, the output end of the first Bottleneck_dcnv2 module, the output end of the second Bottleneck_dcnv2 module and the output end of the third Bottleneck_dcnv2 module are spliced according to a channel and then input to the input end of the second Conv module, and the output end of the second Conv module is used as the output end of the C2f_dcnv2 module;
step 2.2), for the head network of the YOLOv8 model, replacing all the C2f modules with c2f_dcnv2 modules to obtain the YOLOv8 model which introduces deformable convolution;
step 2.3), for the backbone network of the YOLOv8 model introduced with the deformable convolution, replacing all Conv modules except the first Conv module with the receptive field attention convolution module RFAConv to obtain an improved YOLOv8 model;
step 3), training the improved YOLOv8 model based on a training set, a verification set and a test set to obtain a trained improved YOLOv8 model;
and 4) inputting the pavement disease image to be detected into the trained improved YOLOv8 model for recognition, and obtaining a detection result.
As a further optimization scheme of the pavement disease detection method based on the improved YOLOv8 model, in the step 1), the images in the pavement disease image dataset are all marked with pavement disease frames by using labelme marking software, and marking information comprises disease types and coordinates.
As a further optimization scheme of the pavement disease detection method based on the improved YOLOv8 model, the pavement disease image data set in the step 1) is as follows 8:1: the scale of 1 is divided into a training set, a validation set and a test set.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
1. according to the invention, sample equalization and sample amplification are carried out on the data set, so that the generalization capability of the model can be improved, and the model is prevented from being over-fitted;
2. the invention introduces deformable convolution, can adaptively adjust and adapt to irregular disease shapes, and better captures local characteristics;
3. the invention adds a receptive field attention mechanism, can strengthen the receptive field of the neural network, adapt to pavement disease detection with different dimensions and overcome the interference of complex background.
Drawings
FIG. 1 is a schematic flow chart of the present invention;
FIG. 2 is a schematic representation of a pavement slab image in accordance with the present invention;
FIG. 3 is a schematic diagram showing the comparison of road images before and after data amplification in the present invention;
FIG. 4 is a schematic structural diagram of C2f_dcnv2 in the present invention;
FIG. 5 is a schematic diagram of the structure of a deformable convolution DCNv2 in accordance with the present invention;
FIG. 6 is a schematic diagram of the construction of the receptive field attention convolution module RFAConv of the present invention;
FIG. 7 is a schematic diagram of the overall network architecture of the present invention;
fig. 8 is a schematic diagram of the road disease detection result in the present invention.
Detailed Description
The technical scheme of the invention is further described in detail below with reference to the accompanying drawings:
this invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. In the drawings, the components are exaggerated for clarity.
The main application scene of the invention is as follows: under the influence of factors such as extreme environment and traffic bearing capacity, various damages and defects appear on the road, and road diseases need to be detected in the process of road maintenance so as to repair the road, thereby ensuring the safety and durability of the road. Therefore, the invention mainly works as the research of the road disease detection algorithm based on deep learning and the optimization of the detection algorithm, so that the road disease detection algorithm can more efficiently detect the defects of different sizes and shapes in road diseases and has certain robustness.
Referring to fig. 1, a flowchart of a pavement disease detection method based on an improved YOLOv8 model according to an embodiment of the present invention is shown, and the method includes the following steps:
and 1) acquiring a pavement disease image data set and dividing the pavement disease image data set into a training set, a verification set and a test set.
In step 1), the pavement disease image dataset may be a public dataset or may be acquired by acquisition, and it is composed of a preset number of pavement disease images containing labeling information, and is used for improving the training phase of the YOLOv8 target detection model.
When the road surface disease image data set is acquired through acquisition, the method comprises the following steps of:
step 1.1), acquiring a preset number of pavement images, screening pavement images containing pavement diseases, and using labelme marking software to mark 7 diseases such as cracks, crack patches, fissures, crack patches, pits, pit patches and well covers in a frame manner to obtain pavement disease images containing marking information, wherein the marking information contains disease types and coordinates, as shown in figure 2;
step 1.2), aiming at the problem of sample imbalance in the data set, undersampling the categories with more samples, and reducing the number of samples; image enhancement is carried out on categories with fewer samples, including methods of random addition of salt and pepper noise, random histogram equalization, random adjustment of saturation, brightness, contrast, sharpness and the like of images, image changes under different illumination and shooting conditions are simulated, so that sample amplification is realized while sample category information is maintained, and the image after image enhancement is shown in fig. 3;
step 1.3), data sets were assembled according to 8:1: the 1 scale is divided into a training set, a verification set and a test set.
To this end, a data set for the training of the road surface disease detection model can be obtained.
Step 2), constructing an improved YOLOv8 model which introduces a deformable convolution and receptive field attention mechanism.
After the data set for training is acquired, a disease detection network is required to be constructed, and it is required to be explained that the disease detection network is a YOLOv8 network, but the detection effect of the original YOLOv8 model on road surface diseases is not very ideal, and the detection precision is low, so that the disease detection network is required to be further improved, and the road surface diseases can be detected more accurately.
Step 2.1), constructing a C2f_dcnv2 module;
for the header network of the YOLOv8 model, the c2f_dcnv2 module is used to replace all C2f modules; the shape and size of the road fault may vary depending on the road surface condition and maintenance conditions, and the deformable convolution may adaptively adjust the shape of the convolution kernel to better capture and identify various irregular faults, such as cracks, pits, and the like.
Referring to a schematic structural diagram of a c2f_dcnv2 module shown in fig. 4, the c2f_dcnv2 module includes first to second Conv modules, split modules, and first to third bottleneck_dcnv2 modules;
the first to third Bottleneck_dcnv2 modules have the same structure and comprise a third Conv module, a deformable convolution layer DCNv2, a batch normalization layer and a SiLU activation function layer which are sequentially connected, wherein the input end of the third Conv module is connected with external input, and the input end of the third Conv module and the output end of the SiLU activation function layer are added and then used as output;
the first Conv module, the Split module, the first Bottleneck_dcnv2 module, the second Bottleneck_dcnv2 module and the third Bottleneck_dcnv2 module are sequentially connected, the input end of the first Conv module is used as the input end of the C2f_dcnv2 module, the output end of the Split module, the output end of the first Bottleneck_dcnv2 module, the output end of the second Bottleneck_dcnv2 module and the output end of the third Bottleneck_dcnv2 module are spliced according to a channel and then input to the input end of the second Conv module, and the output end of the second Conv module is used as the output end of the C2f_dcnv2 module.
Referring to the schematic diagram of the deformable convolution DCNv2 shown in fig. 5, the deformable convolution layer obtains a set of prediction results of convolution kernel offset and a set of prediction results of weight through a convolution operation acting on an input feature map, the size of the offset feature map is consistent with that of the input feature map, the number of channels is 2N, wherein 2 means that each offset is (x, y) two values, and N is the number of pixels of the convolution kernel, i.e. the deformable convolution learns a set of offset for each pixel of the input feature map. The DCNv2 adopted in the embodiment introduces a modulation mechanism, utilizes weights to adjust the input characteristic amplitude values from different space positions, and has the same size as the input characteristic graph and N channels;
given K sampling positionsLet w be k And p k The weight and the pre-specified offset of the kth position are respectively represented, and x (p) and y (p) are respectively represented as features of the position p in the input feature map x and the output feature map y. The expression of the deformable convolution DCNv2 is shown in the formula (1);
wherein Δp k And Δm k The leachable offset and the weight coefficient for the kth position, respectively. Weight coefficient Deltam k At [0,1]Within a range Δp k As real numbers, the range is not constrained. Δp k And Δm k Respectively initializing to 0 and 0.5 (default offset is 0, sampling point contribution rate cannot be distinguished), corresponding convolution kernel parameters are initially 0, and the learning rate of a convolution layer is set to be 0.1 of an existing layer.
Step 2.2), for the head network of the YOLOv8 model, replacing all C2f modules with c2f_dcnv2 modules, resulting in a YOLOv8 model that introduces a deformable convolution.
Step 2.3) for the backbone network introducing the deformable convolutions YOLOv8 model, replacing all Conv modules except the first Conv module with the receptive field attention convolution module RFAConv, resulting in an improved YOLOv8 model.
Spatial attention is widely applied to improving the performance of convolutional neural networks, but the existing spatial attention, such as a convolutional block attention module CBAM and a coordinated attention CA, cannot effectively solve the problem of parameter sharing of large-scale convolution kernels (such as 3 x 3 convolution), because the spatial features of the whole receptive field are not considered, the receptive field attention RFA uses non-overlapping receptive field slider extraction features, the importance of various features in the receptive field slider is emphasized, and the spatial features of the receptive field are also focused, so that the problem of parameter sharing of the convolution kernels is thoroughly solved.
The receptive field attention convolution module RFAConv is shown in fig. 6, in order to ensure small calculation cost and parameter quantity, RFAConv uses AvgPool to aggregate global features of each receptive field feature, then uses 1×1 group convolution to perform information interaction, and finally uses softmax to emphasize importance of each feature in the receptive field feature, and RFA is calculated as shown in formula (2);
in the formula g 1×1 Representing a block convolution of size 1X 1, k representing the size of the convolution kernel, k=3 selected, norm representing normalization, X representing the input feature map, and F represented by the attention map a rf And transformed receptive field spatial features F rf Multiplication.
In step 2), an existing YOLOv8 model is modified, and the modified YOLOv8 pavement disease detection model is shown with reference to fig. 7. The head network of the YOLOv8 model uses the C2f_dcnv2 module to replace all the C2f modules, the shape and the size of road diseases can be different according to the pavement condition and the maintenance condition, and the shape of the convolution kernel can be adaptively adjusted by the deformable convolution so as to better capture and identify various irregular diseases such as cracks, pits and the like. For a main network of the YOLOv8 model introduced with deformable convolution, all Conv modules except the first Conv module are replaced by a receptive field attention convolution module RFAConv, receptive fields of the neural network are enhanced, and the improved YOLOv8 model is obtained after pavement disease detection of different dimensions is adapted.
And 3) training the improved YOLOv8 model based on the training set, the verification set and the test set to obtain the trained improved YOLOv8 model.
Step 3.1), in the training stage, firstly initializing model parameters of an improved YOLOv8 target detection model;
step 3.2), inputting the pavement defect image in the pavement defect image data set into the improved YOLOv8 target detection model for recognition, and obtaining a detection frame and category information; inputting the detection frame, the category information and the marking information of the pavement defect image into a preset optimizing Loss function, and calculating a Loss value, wherein the Loss function is frame regression Loss CIoU Loss and DFL Loss, and category classification cross entropy Loss;
and 3.3) setting a gradient descent algorithm as SGD, optimizing model parameters of the improved YOLOv8 target detection model, and setting a training round as 650epochs to obtain the trained improved YOLOv8 target detection model.
The superiority of the improved road surface disease detection network was evaluated by Precision, recall and average accuracy, and the training results of the improved road surface disease detection network were compared with the original network which was not improved, and the results indicate that the improved YOLOv8 network exhibited better accuracy in road surface disease detection, and the results are shown in tables 1 and 2, wherein table 1 represents the data of the original YOLOv8 detection, and table 2 represents the data of the improved YOLOv8 network.
TABLE 1 detection results of original YOLOv8 model
TABLE 2 improved YOLOv8 model test results
Step 4), inputting the pavement disease image to be detected into a trained improved YOLOv8 model for recognition, and obtaining a detection result;
inputting the road surface image to be detected into a trained optimized disease detection model, and outputting a road surface disease detection image shown in fig. 8; the disease detection image comprises disease categories of each disease, and the disease detection image is marked by a rectangular frame and confidence is marked.
In summary, according to the pavement disease detection method based on the improved YOLOv8 model provided by the embodiment of the invention, firstly, a pavement disease data set is obtained, and secondly, a C2f_dcnv2 module is used for replacing all C2f modules of the original YOLOv8 model head network, so that various irregular diseases such as cracks, pits and the like are better captured and identified, and the detection precision is improved; furthermore, a receptive field attention convolution module RFAConv is used for replacing all Conv modules except a first Conv module of a main network of an original YOLOv8 model, so that the feature extraction capacity of the model is improved, the receptive field of a neural network is enhanced, and the method is suitable for pavement disease detection of different scales; then, training the improved model to obtain a model with higher detection precision; finally, the model obtained through training is used for detecting the pavement image, so that a final detection effect can be obtained.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
While the foregoing is directed to embodiments of the present invention, other and further details of the invention may be had by the present invention, it should be understood that the foregoing description is merely illustrative of the present invention and that no limitations are intended to the scope of the invention, except insofar as modifications, equivalents, improvements or modifications are within the spirit and principles of the invention.
Claims (3)
1. The pavement disease detection method based on the improved YOLOv8 model is characterized by comprising the following steps of:
step 1), obtaining a pavement disease image data set and dividing the pavement disease image data set into a training set, a verification set and a test set;
step 2), constructing an improved YOLOv8 model which introduces a deformable convolution and receptive field attention mechanism;
step 2.1), constructing a C2f_dcnv2 module;
the C2f_dcnv2 module comprises a first Conv module, a second Conv module, a Split module and a first Bottleneck_dcnv2 module;
the first to third Bottleneck_dcnv2 modules have the same structure and comprise a third Conv module, a deformable convolution layer DCNv2, a batch normalization layer and a SiLU activation function layer which are sequentially connected, wherein the input end of the third Conv module is connected with external input, and the input end of the third Conv module and the output end of the SiLU activation function layer are added and then used as output;
the first Conv module, the Split module, the first Bottleneck_dcnv2 module, the second Bottleneck_dcnv2 module and the third Bottleneck_dcnv2 module are sequentially connected, the input end of the first Conv module is used as the input end of the C2f_dcnv2 module, the output end of the Split module, the output end of the first Bottleneck_dcnv2 module, the output end of the second Bottleneck_dcnv2 module and the output end of the third Bottleneck_dcnv2 module are spliced according to a channel and then input to the input end of the second Conv module, and the output end of the second Conv module is used as the output end of the C2f_dcnv2 module;
step 2.2), for the head network of the YOLOv8 model, replacing all the C2f modules with c2f_dcnv2 modules to obtain the YOLOv8 model which introduces deformable convolution;
step 2.3), for the backbone network of the YOLOv8 model introduced with the deformable convolution, replacing all Conv modules except the first Conv module with the receptive field attention convolution module RFAConv to obtain an improved YOLOv8 model;
step 3), training the improved YOLOv8 model based on a training set, a verification set and a test set to obtain a trained improved YOLOv8 model;
and 4) inputting the pavement disease image to be detected into the trained improved YOLOv8 model for recognition, and obtaining a detection result.
2. The pavement disease detection method based on the improved YOLOv8 model according to claim 1, wherein in the step 1), the images in the pavement disease image dataset are all marked by using labelme marking software, and marking information comprises disease types and coordinates.
3. The method for detecting road surface diseases based on the modified YOLOv8 model according to claim 1, wherein the road surface disease image dataset in step 1) is according to 8:1: the scale of 1 is divided into a training set, a validation set and a test set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410021694.0A CN117788950A (en) | 2024-01-05 | 2024-01-05 | Pavement disease detection method based on improved YOLOv8 model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410021694.0A CN117788950A (en) | 2024-01-05 | 2024-01-05 | Pavement disease detection method based on improved YOLOv8 model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117788950A true CN117788950A (en) | 2024-03-29 |
Family
ID=90381622
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410021694.0A Pending CN117788950A (en) | 2024-01-05 | 2024-01-05 | Pavement disease detection method based on improved YOLOv8 model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117788950A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117975401A (en) * | 2024-04-02 | 2024-05-03 | 深圳市锐明像素科技有限公司 | Road disease detection method, device, electronic equipment and storage medium |
-
2024
- 2024-01-05 CN CN202410021694.0A patent/CN117788950A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117975401A (en) * | 2024-04-02 | 2024-05-03 | 深圳市锐明像素科技有限公司 | Road disease detection method, device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113469953B (en) | Transmission line insulator defect detection method based on improved YOLOv4 algorithm | |
| CN110070008B (en) | Bridge disease identification method adopting unmanned aerial vehicle image | |
| CN110378232B (en) | Improved test room examinee position rapid detection method of SSD dual-network | |
| CN111784633A (en) | Insulator defect automatic detection algorithm for power inspection video | |
| CN114973002A (en) | Improved YOLOv 5-based ear detection method | |
| CN116310785B (en) | Unmanned aerial vehicle image pavement disease detection method based on YOLO v4 | |
| CN109840904B (en) | Detection method for large-scale difference parts of high-speed rail contact network | |
| CN113327255A (en) | Power transmission line inspection image processing method based on YOLOv3 detection, positioning and cutting and fine-tune | |
| CN110321944A (en) | A kind of construction method of the deep neural network model based on contact net image quality evaluation | |
| CN113256602A (en) | Unsupervised fan blade defect detection method and system based on self-encoder | |
| CN112489026A (en) | Asphalt pavement disease detection method based on multi-branch parallel convolution neural network | |
| CN109509188A (en) | A kind of transmission line of electricity typical defect recognition methods based on HOG feature | |
| CN113469097B (en) | Multi-camera real-time detection method for water surface floaters based on SSD network | |
| CN111291818A (en) | Non-uniform class sample equalization method for cloud mask | |
| CN110618129A (en) | Automatic power grid wire clamp detection and defect identification method and device | |
| CN112734732B (en) | Railway tunnel cable leakage clamp detection method based on improved SSD algorithm | |
| CN112132088B (en) | Inspection point missing inspection identification method | |
| CN113450318A (en) | Primary power distribution system porcelain insulator detection method based on unmanned aerial vehicle autonomous vision | |
| CN113762144A (en) | Deep learning-based black smoke vehicle detection method | |
| CN117788950A (en) | Pavement disease detection method based on improved YOLOv8 model | |
| CN116958073A (en) | Small sample steel defect detection method based on attention feature pyramid mechanism | |
| CN116129327A (en) | Infrared vehicle detection method based on improved YOLOv7 algorithm | |
| CN114463678A (en) | Rainfall type identification method using camera video image | |
| CN116485802B (en) | Insulator flashover defect detection method, device, equipment and storage medium | |
| CN113947567B (en) | Defect detection method based on multitask learning |
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 |