CN114445712A - Expressway pavement disease identification method based on improved YOLOv5 model - Google Patents

Expressway pavement disease identification method based on improved YOLOv5 model Download PDF

Info

Publication number
CN114445712A
CN114445712A CN202210113377.2A CN202210113377A CN114445712A CN 114445712 A CN114445712 A CN 114445712A CN 202210113377 A CN202210113377 A CN 202210113377A CN 114445712 A CN114445712 A CN 114445712A
Authority
CN
China
Prior art keywords
layer
model
yolov5
steps
yolov5 model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210113377.2A
Other languages
Chinese (zh)
Inventor
赵池航
覃晓明
毛迎兵
刘洋
吴加伦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Original Assignee
Southeast University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN202210113377.2A priority Critical patent/CN114445712A/en
Publication of CN114445712A publication Critical patent/CN114445712A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an expressway pavement disease identification method based on an improved YOLOv5 model, which comprises the following steps: on the basis of the existing mature YOLOv5 model, a target detection model suitable for highway pavement disease identification is constructed by improving a model feature extraction network and redesigning an anchor frame (anchor) of the model. By the close combination of the deep learning network, the invention can greatly improve the efficiency of disease identification and provide technical support for highway maintenance when being applied to the field of highway pavement disease identification.

Description

Expressway pavement disease identification method based on improved YOLOv5 model
Technical Field
The invention relates to the field of intelligent traffic and intelligent high-speed research, in particular to an expressway pavement disease identification method based on an improved YOLOv5 model.
Background
With the increasing traffic flow on expressways and the existence of some illegal and overloaded vehicles, various pavement diseases begin to appear on many expressways, the traditional artificial-based pavement disease identification can not meet the requirements of maintenance of a large number of expressways, and a new solution and thought are provided for full-automatic and rapid acquisition and identification of the pavement diseases of the expressways by realizing full-automatic acquisition of pavement information of the expressways based on a pavement detection vehicle and a target detection algorithm based on deep learning and computer vision. In view of the above, the method for identifying the highway pavement diseases based on the improved YOLOv5 is researched by constructing the YOLOv5 model of different feature extraction networks on the basis of analyzing the overall frame structure of the YOLOv5 model by performing deep learning modeling on pavement disease image data.
Disclosure of Invention
The purpose of the invention is as follows: in order to overcome the defects in the prior art, the provided expressway pavement disease identification method based on deep learning utilizes the deep learning method based on the improved YOLOv5 model to quickly and effectively identify and classify expressway pavement diseases, and can provide technical support for expressway pavement maintenance.
The technical scheme is as follows: in order to achieve the purpose, the method for identifying the highway pavement diseases based on the improved YOLOv5 provided by the invention comprises the following steps:
s1: constructing a YOLOv5 target detection model based on different feature extraction networks, and performing feature extraction on the highway pavement image;
s2, redesign of the anchor frame (anchor) of the model.
Further, a YOLOv5 target detection model based on different feature extraction networks is constructed in the step S1, and feature extraction is performed on the highway pavement image:
s1-1: constructing an Efficientnet-YOLOv5 model;
s1-2: constructing a Mobilenet v3-YOLOv5 model;
s1-3: constructing a Resnet50-YOLOv5 model;
further, the method for constructing the Efficientnet-Yolov5 model comprises the following steps: in an EfficientNet-B0 network, the resolution of an input image is 224 multiplied by 224, convolution operation is carried out in a first Stage layer through a convolution kernel with the size of 3 multiplied by 3, and the number of channels of an output feature map is increased to 32; extracting image features in the second Stage layer to the eighth Stage layer by continuously superposing MBConv structures, continuously reducing the size of a feature map in the process, and increasing the channel number of the feature map to extract features of higher layers; in the ninth Stage layer, a point-by-point convolution layer, a maximum pooling layer and a full-link layer are used to be connected in sequence. The feature extraction network is brought into a YOLOv5 model to obtain an Efficientnet-YOLOv5 model.
Further, the method for constructing the Mobilene v3-YOLOv5 model comprises the following steps: the Mobilenetv3 network firstly carries out a conventional convolution operation on the input image, so that the number of channels of the output feature map is 16; then extracting image features through 15 bneck modules, gradually reducing the size of a feature map, and increasing the number of channels of the feature map, so that the level of feature map extraction information is improved, the number of parameters of the models is reduced on the premise of ensuring the quality of feature extraction, and all the bneck modules are connected by using residual errors; then, information among the channels is interacted through point-by-point convolution, and the number of the channels of the feature diagram is increased to 960; then extracting a feature vector from the 7 multiplied by 7 feature map through a pooling operation; and finally, the number of channels of the output feature graph is increased through two 1 × 1 point-by-point convolution layers, and features of higher layers are extracted. The feature extraction network is brought into a YOLOv5 model to obtain a Mobilenet v3-YOLOv5 model.
Further, the method for constructing the Resnet50-Yolov5 model comprises the following steps: the Resnet50 network consists of conv1 layer, conv2_ x layer, conv3_ x layer, conv4_ x layer, conv5_ x layer, as well as a full connection layer and a softmax classifier. In the whole network structure, the conv1 layer contains a convolution layer with 7 × 7 convolution kernel step size of 2, the conv2_ x layer contains 3 repeated 3-layer residual units, the conv3_ x layer contains 4 repeated 3-layer residual units, the conv4_ x layer contains 6 repeated 3-layer residual units, and the conv5_ x layer contains 3 repeated 3-layer residual units. And (4) bringing the feature extraction network into a YOLOv5 model to obtain a Resnet50-YOLOv5 model.
Further, the step S2 redesigns the anchor frame (anchor) of the model, and includes the following steps:
s2-1: collecting all the labeled bounding box length and width data in the labeled data set, and clustering by using a Kmeans clustering method;
s2-2: the number of Kmeans cluster center points is set to 9. Obtaining length and width parameters of 9 anchor frames (anchors) after clustering;
the method improves the feature extraction network in the YOLOv5, increases the network depth, uses Mosaic data enhancement on the image and carries out self-adaptive anchor frame (anchor) calculation.
Has the advantages that: compared with the prior art, the method is improved on the basis of the original deep learning network, so that the detection and classification of the diseases can be more accurately realized, and the technical support is provided for the pavement maintenance of the expressway.
Drawings
FIG. 1 is a comparison of a bar chart and a box chart of the F1-score index for the four models after feature network improvement, where a is the bar chart comparison and b is the box chart comparison.
FIG. 2 is a comparison of the bars and boxes for the F1-score index for the three models after the improvement of the anchor box, where a is the bar comparison and b is the box comparison.
Fig. 3 is a YOLOv5 network structure.
Detailed Description
The present invention will be further illustrated with reference to the accompanying drawings and specific embodiments, which are to be understood as merely illustrative of the invention and not as limiting the scope of the invention.
The invention provides an improved YOLOv 5-based highway pavement disease identification method, which comprises the following steps:
s1: constructing a YOLOv5 target detection model based on different feature extraction networks, and performing feature extraction on the highway pavement image:
s1-1: an Efficientnet-Yolov5 model was constructed.
The resolution of an input image of an EfficientNet network (shown in Table 1) is 224 multiplied by 224, convolution operation is carried out in a first Stage layer through a convolution kernel with the size of 3 multiplied by 3, and the number of channels of an output feature map is increased to 32; extracting image features in the second Stage layer to the eighth Stage layer by continuously superposing MBConv structures, continuously reducing the size of a feature map in the process, and increasing the channel number of the feature map to extract features of higher layers; in the ninth Stage layer, a point-by-point convolution layer, a maximum pooling layer and a full-link layer are used to be connected in sequence. The feature extraction network is brought into a YOLOv5 model to obtain an Efficientnet-YOLOv5 model.
TABLE 1 EfficientNet-B0 network structure Table
Stage Operator Resolution Channels Layers
1 conv3×3 224×224 32 1
2 MBConv1，k3×3 112×112 16 1
3 MBConv1，k33 112×112 24 2
4 MBConv1，k33 56×56 40 2
5 MBConv1.k33 28×28 8() 3
6 MBConv1，k33 14×14 112 3
7 MBConv1，k33 14×14 192 4
8 MBConv1，k33 7×7 320 1
9 conv1×1&pool&FC 7×7 1280 1
S1-2: a Mobilenet v3-YOLOv5 model was constructed.
The Mobilenet v3 network (shown in table 2) first performs a conventional convolution operation on the input image, so that the number of channels of the output feature map is 16; then extracting image features through 15 bneck modules, gradually reducing the size of the feature map, and increasing the number of channels of the feature map, so that the level of feature map extraction information is improved, the number of parameters of the models is reduced on the premise of ensuring the feature extraction quality, and the bneck modules are connected by using residual errors; then, information among the channels is interacted through point-by-point convolution, and the number of the channels of the feature diagram is increased to 960; then extracting a feature vector from the 7 multiplied by 7 feature map through a pooling operation; and finally, the number of channels of the output feature graph is increased through two 1 × 1 point-by-point convolution layers, and features of higher layers are extracted. And (3) bringing the feature extraction network into a YOLOv5 model to obtain a Mobilenet v3-YOLOv5 model.
TABLE 2 Mobilene v3 network architecture Table
Input Operator exp size out NL s
2242×3 conv2d - 16 HS 2
1122×16 bneck，3×3 16 16 RE 1
1122×16 bneck，3×3 64 24 RE 2
562×24 bneck，3×3 72 24 RE 1
562×24 bneck，5×5 72 40 RE 2
282×40 bneck，5×5 120 40 RE 1
282×40 bneck，5×5 120 40 RE 1
282×40 bneck，3×3 240 80 HS 2
142×80 bneck，3×3 200 80 HS 1
142×80 bneck，3×3 184 80 HS 1
142×80 bneck，3×3 184 80 HS 1
142×80 bneck，3×3 480 112 HS 1
142×112 bneck，3×3 672 112 HS 1
142×112 bneck，5×5 672 160 HS 2
72×160 bneck，5×5 960 160 HS 1
72×160 bneck，5×5 960 160 HS 1
72×160 conv2d，1×1 - 960 HS 1
72×960 pool，7×7 - - 1
12×960 conv2d 1×1，NBN - 1280 HS 1
12×1280 conv2d 1×1，NBN - k - 1
S1-3: the Resnet50-YOLOv5 model was constructed.
The Resnet50 network (shown in Table 3) consists of conv1 layer, conv2_ x layer, conv3_ x layer, conv4_ xlayer, conv5_ x layer, a full link layer and a softmax classifier. In the whole network structure, the conv1 layer contains a convolution layer with 7 × 7 convolution kernel step size of 2, the conv2_ x layer contains 3 repeated 3-layer residual units, the conv3_ x layer contains 4 repeated 3-layer residual units, the conv4_ x layer contains 6 repeated 3-layer residual units, and the conv5_ x layer contains 3 repeated 3-layer residual units. And (4) bringing the feature extraction network into a YOLOv5 model to obtain a Resnet50-YOLOv5 model.
Table 3 Resnet network structure table
Figure BDA0003494926830000051
S2: redesigning the anchor frame (anchor) of the model:
s2-1: collecting all the labeled bounding box length and width data in the labeled data set, and clustering by using a Kmeans clustering method;
s2-2: setting the Kmeans clustering central points to be 9, and obtaining length and width parameters of 9 anchor frames (anchors) after clustering;
in order to verify the effect of the above method, in this embodiment, experimental comparison is performed by modifying backsbone (feature extraction network) in the YOLOv5 target detection model of step S1 to EfficientNet, Resnet50, and mobilent v3, as shown in table 4 and fig. 1.
TABLE 4 test results of four models
Figure BDA0003494926830000061
It can be seen that the improved Yolov5 network model for identifying highway pavement diseases is superior to the original Yolov5 network model, the accuracy rate reaches 0.8449, the recall rate is 88.02%, and the F1-score index can reach 0.8622. According to S2, removing YOLOv5-Resnet50 with poor experimental results, modifying anchor frame (anchor) parameters in three models of YOLOv5 model, YOLOv5-Efficientnet model and YOLOv5-Mobilenetv3 model, and performing the experiment again, as shown in Table 5 and FIG. 2:
TABLE 5 test results of three models after redesigning anchor frame
Figure BDA0003494926830000062
It can be seen that the three models have obvious improvement on three evaluation indexes of Precision (Precision), Recall (Recall) and harmonic mean (F1-score), wherein the improvement of the Recall (Recall) of the Yolov 5-efficiency internet model is most obvious, and the Recall value reaches 91.26% after the anchor frame (anchor) is redesigned. The evaluation index value of the YOLOv5 model F1-score is improved by about 2%, and the evaluation index values of the YOLOv5-Efficientnet model and the YOLOv5-Mobilene v3 model F1-score are improved by nearly 2.5% in a similar way.
The technical means disclosed in the invention scheme are not limited to the technical means disclosed in the above embodiments, but also include the technical scheme formed by any combination of the above technical features. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and such improvements and modifications are also considered to be within the scope of the present invention.

Claims (6)

1. The highway pavement disease identification method based on the improved YOLOv5 model is characterized by comprising the following steps: the method comprises the following steps:
s1: constructing a YOLOv5 target detection model based on different feature extraction networks, and performing feature extraction on the highway pavement image;
s2, redesigning the anchor frame of the model.
2. The method for identifying the pavement diseases of the expressway based on the improved YOLOv5 model according to claim 1, wherein the method comprises the following steps: the specific steps of redesigning the anchor frame of the model in step S2 are as follows:
the method comprises the following steps: collecting all the labeled bounding box length and width data in the labeled data set, and clustering by using a Kmeans clustering method;
secondly, the step of: the number of Kmeans clustering center points is set to be 9, and the length and width parameters of 9 anchor frames (anchors) are obtained after clustering.
3. The method for identifying the pavement diseases of the expressway based on the improved YOLOv5 model according to claim 1, wherein the method comprises the following steps: the method for constructing the YOLOv5 target detection model based on different feature extraction networks in the step S1 includes: an Efficientnet network, a Mobilenet v3 and a Resnet50 are taken as feature extraction networks to be brought into a YOLOv5 model, and an Efficientnet-YOLOv5 model, a Mobilenet v2-YOLOv5 model and a Resnet50-YOLOv5 model are respectively constructed.
4. The method for identifying the pavement diseases of the expressway based on the improved YOLOv5 model according to claim 3, wherein the method comprises the following steps: the method for constructing the Efficientnet-Yolov5 model comprises the following steps: in an EfficientNet-B0 network, the resolution of an input image is 224 multiplied by 224, convolution operation is carried out in a first Stage layer through a convolution kernel with the size of 3 multiplied by 3, and the number of channels of an output feature map is increased to 32; extracting image features in the second Stage layer to the eighth Stage layer by continuously superposing MBConv structures, continuously reducing the size of a feature map in the process, and increasing the channel number of the feature map to extract features of higher layers; in the ninth Stage layer, a point-by-point convolution layer, a maximum pooling layer and a full-connection layer are sequentially connected; and bringing the feature extraction network into a YOLOv5 model to obtain an Efficientnet-YOLOv5 model.
5. The method for identifying the pavement diseases of the expressway based on the improved YOLOv5 model according to claim 3, wherein the method comprises the following steps: the method for constructing the Mobilenet v3-YOLOv5 model comprises the following steps: the Mobilenetv3 network firstly carries out a conventional convolution operation on the input image, so that the number of channels of the output feature map is 16; then extracting image features through 15 bneck modules, gradually reducing the size of a feature map, and increasing the number of channels of the feature map, so that the level of feature map extraction information is improved, the number of parameters of the models is reduced on the premise of ensuring the quality of feature extraction, and all the bneck modules are connected by using residual errors; then, information among the channels is interacted through point-by-point convolution, and the number of the channels of the feature diagram is increased to 960; then extracting a feature vector from the 7 multiplied by 7 feature map through a pooling operation; finally, the number of channels of the output feature graph is increased through two 1 x 1 point-by-point convolution layers, and features of higher layers are extracted; the feature extraction network is brought into a YOLOv5 model to obtain a Mobilenet v3-YOLOv5 model.
6. The method for identifying the pavement diseases of the expressway based on the improved YOLOv5 model according to claim 3, wherein the method comprises the following steps: the method for constructing the Resnet50-YOLOv5 model comprises the following steps: the Resnet50 network consists of a conv1 layer, a conv2_ x layer, a conv3_ x layer, a conv4_ x layer, a conv5_ x layer, a full connection layer and a softmax classifier; in the whole network structure, a conv1 layer comprises a convolution layer with a convolution kernel step size of 7 × 7 being 2, a conv2_ x layer comprises 3 repeated 3-layer residual units, a conv3_ x layer comprises 4 repeated 3-layer residual units, a conv4_ x layer comprises 6 repeated 3-layer residual units, and a conv5_ x layer comprises 3 repeated 3-layer residual units; and (4) bringing the feature extraction network into a YOLOv5 model to obtain a Resnet50-YOLOv5 model.
CN202210113377.2A 2022-01-29 2022-01-29 Expressway pavement disease identification method based on improved YOLOv5 model Pending CN114445712A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210113377.2A CN114445712A (en) 2022-01-29 2022-01-29 Expressway pavement disease identification method based on improved YOLOv5 model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210113377.2A CN114445712A (en) 2022-01-29 2022-01-29 Expressway pavement disease identification method based on improved YOLOv5 model

Publications (1)

Publication Number Publication Date
CN114445712A true CN114445712A (en) 2022-05-06

Family

ID=81372269

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210113377.2A Pending CN114445712A (en) 2022-01-29 2022-01-29 Expressway pavement disease identification method based on improved YOLOv5 model

Country Status (1)

Country Link
CN (1) CN114445712A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116229396A (en) * 2023-02-17 2023-06-06 广州丰石科技有限公司 High-speed pavement disease identification and warning method

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111652171A (en) * 2020-06-09 2020-09-11 电子科技大学 Construction method of facial expression recognition model based on double branch network
CN111986188A (en) * 2020-08-27 2020-11-24 深圳市智源空间创新科技有限公司 Capsule robot drainage pipe network defect identification method based on Resnet and LSTM
CN113205107A (en) * 2020-11-02 2021-08-03 哈尔滨理工大学 Vehicle type recognition method based on improved high-efficiency network
CN113256601A (en) * 2021-06-10 2021-08-13 北方民族大学 Pavement disease detection method and system
CN113537244A (en) * 2021-07-23 2021-10-22 深圳职业技术学院 Livestock image target detection method and device based on light-weight YOLOv4
CN113609911A (en) * 2021-07-07 2021-11-05 北京工业大学 Pavement disease automatic detection method and system based on deep learning

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111652171A (en) * 2020-06-09 2020-09-11 电子科技大学 Construction method of facial expression recognition model based on double branch network
CN111986188A (en) * 2020-08-27 2020-11-24 深圳市智源空间创新科技有限公司 Capsule robot drainage pipe network defect identification method based on Resnet and LSTM
CN113205107A (en) * 2020-11-02 2021-08-03 哈尔滨理工大学 Vehicle type recognition method based on improved high-efficiency network
CN113256601A (en) * 2021-06-10 2021-08-13 北方民族大学 Pavement disease detection method and system
CN113609911A (en) * 2021-07-07 2021-11-05 北京工业大学 Pavement disease automatic detection method and system based on deep learning
CN113537244A (en) * 2021-07-23 2021-10-22 深圳职业技术学院 Livestock image target detection method and device based on light-weight YOLOv4

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116229396A (en) * 2023-02-17 2023-06-06 广州丰石科技有限公司 High-speed pavement disease identification and warning method
CN116229396B (en) * 2023-02-17 2023-11-03 广州丰石科技有限公司 High-speed pavement disease identification and warning method

Similar Documents

Publication Publication Date Title
CN111259905B (en) Feature fusion remote sensing image semantic segmentation method based on downsampling
CN108876780B (en) Bridge crack image crack detection method under complex background
CN105335716B (en) A kind of pedestrian detection method extracting union feature based on improvement UDN
CN112183667B (en) Insulator fault detection method in cooperation with deep learning
CN106557579B (en) Vehicle model retrieval system and method based on convolutional neural network
CN110717147A (en) Method for constructing driving condition of automobile
CN110032952B (en) Road boundary point detection method based on deep learning
CN110349170B (en) Full-connection CRF cascade FCN and K mean brain tumor segmentation algorithm
CN111105389A (en) Detection method for pavement crack by fusing Gabor filter and convolutional neural network
CN115049640B (en) Road crack detection method based on deep learning
CN111882620A (en) Road drivable area segmentation method based on multi-scale information
CN114445712A (en) Expressway pavement disease identification method based on improved YOLOv5 model
CN114782949B (en) Traffic scene semantic segmentation method for boundary guide context aggregation
CN112766283A (en) Two-phase flow pattern identification method based on multi-scale convolution network
CN116824543A (en) Automatic driving target detection method based on OD-YOLO
CN116503336A (en) Pavement crack detection method based on deep learning
CN115170479A (en) Automatic extraction method for asphalt pavement repairing diseases
CN115410024A (en) Power image defect detection method based on dynamic activation thermodynamic diagram
CN116704350B (en) Water area change monitoring method and system based on high-resolution remote sensing image and electronic equipment
CN111612803B (en) Vehicle image semantic segmentation method based on image definition
CN109190451B (en) Remote sensing image vehicle detection method based on LFP characteristics
CN116129327A (en) Infrared vehicle detection method based on improved YOLOv7 algorithm
CN113192076B (en) MRI brain tumor image segmentation method combining classification prediction and multi-scale feature extraction
CN112949771A (en) Hyperspectral remote sensing image classification method based on multi-depth multi-scale hierarchical attention fusion mechanism
CN112801102A (en) Network model and detection method for road surface block disease detection

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