CN111507296A - Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning - Google Patents
Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning Download PDFInfo
- Publication number
- CN111507296A CN111507296A CN202010326763.0A CN202010326763A CN111507296A CN 111507296 A CN111507296 A CN 111507296A CN 202010326763 A CN202010326763 A CN 202010326763A CN 111507296 A CN111507296 A CN 111507296A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- aerial vehicle
- unmanned aerial
- vehicle remote
- building
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/176—Urban or other man-made structures
-
- 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
- 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
Abstract
The invention discloses an intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning, which comprises the following steps: collecting remote sensing image data of the multi-source multi-scale violation building unmanned aerial vehicle; constructing an unmanned aerial vehicle remote sensing image preprocessing model, preprocessing unmanned aerial vehicle remote sensing image data based on the preprocessing model, and making a sample based on the preprocessed data to obtain a sample library; an NMS (network management system) method is suppressed based on a sliding window and a non-maximum value, and a violation building identification model is established by combining a deep learning network; training and optimizing the illegal building identification model by using sample data in the sample library; and inputting the remote sensing image of the unmanned aerial vehicle to be detected into the preprocessing model, and inputting the preprocessed data into the trained illegal building identification model to obtain the illegal building in the remote sensing image of the unmanned aerial vehicle to be detected. The invention can rapidly and accurately carry out intelligent detection on the illegal building and provides data support for law enforcement personnel.
Description
Technical Field
The invention relates to the technical field of optical remote sensing, in particular to an intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning.
Background
The essence of target detection and identification based on optical remote sensing images is that an algorithm finds out an expected target in the image and extracts the specific attribute of the target. The detection and identification of remote sensing images has very wide and important applications in both civil and military fields, such as natural disaster observation and prevention, crop growth health analysis, air quality detection, combat plan adjustment and precision combat, and the like. Due to the important role of the optical remote sensing image in daily production and life and national defense safety, the target detection and identification technology based on the optical remote sensing image is a research hotspot in the field of target identification at home and abroad. Taking the research of optical remote sensing image information extraction and target identification of an airplane as an example, the traditional algorithm mainly uses the characteristics of an artificially extracted image to identify the target, wherein the characteristics comprise texture characteristics, histogram of gradient directions (HOG), Gabor transformation and the like, and then the characteristics are input into a traditional classifier in the form of characteristic vectors, such as a Support Vector Machine (SVM), AdaBoost, a decision tree and the like to classify. The traditional method is often poor in the aspects of artificial feature extraction, algorithm robustness, displacement, rotation invariance and the like, and meanwhile, the manual selection of features is often dependent on experience or luck; in recent years, as deep learning is deeply researched in the field of pattern recognition, image detection and recognition algorithms based on deep learning are emerging continuously, and the performance in all aspects is superior to that of the traditional manual extraction method and the machine learning method. Therefore, the application of deep learning to optical remote sensing image detection and recognition becomes a possible option.
Unmanned aerial vehicle remote sensing is a very important means for obtaining high-resolution images of cities. The violation buildings and the legal buildings have great similarity in structure and image characteristics, the difficulty in distinguishing the violation buildings and the legal buildings is high in satellite images with relatively low resolution, the unmanned aerial vehicle images have centimeter-level spatial resolution, abundant building detail information can be identified, the abundant visual information is used as the characteristics of deep learning, a neural network is trained, and the identification rate of the violation buildings can be greatly improved.
Therefore, a method for obtaining a proper characteristic through self-learning and rapidly and accurately intelligently detecting the illegal building is needed at present.
Disclosure of Invention
The invention aims to provide an intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning, which aims to solve the problems in the prior art, can quickly and accurately carry out intelligent detection on illegal buildings and provides data support for law enforcement personnel.
In order to achieve the purpose, the invention provides the following scheme: the invention provides an intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning, which comprises the following steps:
collecting remote sensing image data of the multi-source multi-scale violation building unmanned aerial vehicle;
constructing an unmanned aerial vehicle remote sensing image preprocessing model, preprocessing collected unmanned aerial vehicle remote sensing image data based on the unmanned aerial vehicle remote sensing image preprocessing model, and making a sample based on the preprocessed unmanned aerial vehicle remote sensing image data to obtain a sample library;
an NMS (network management system) method is suppressed based on a sliding window and a non-maximum value, and a violation building identification model is established by combining a deep learning network; training and optimizing the illegal building identification model by using sample data in the sample library;
inputting the unmanned aerial vehicle remote sensing image to be detected into an unmanned aerial vehicle remote sensing image preprocessing model for preprocessing, inputting the preprocessed data into a trained violation building identification model, and completing identification, positioning and extraction of violation buildings to obtain the violation buildings in the unmanned aerial vehicle remote sensing image to be detected.
Preferably, an unmanned aerial vehicle remote sensing image preprocessing model is constructed based on a geographic space data abstraction library GDA L raster data model, and image data preprocessed by the unmanned aerial vehicle remote sensing image preprocessing model is tif format data.
Preferably, the process of making the sample comprises:
firstly, intercepting a preprocessed unmanned aerial vehicle remote sensing image to obtain an image block of a violation building;
secondly, labeling the image block based on the illegal building data sketched in the universe;
and thirdly, performing rotation expansion on the image block in a plurality of directions.
Preferably, the XM L file of the sample data is used to record the label information of the sample image and the image position of each building object in the sample image.
Preferably, the sample library is divided into a training set, a testing set and a cross validation set.
Preferably, the construction and training of the illegal building identification model specifically comprises the following steps:
firstly, constructing a violation building identification model based on a sliding window and non-maximum value suppression NMS method and combining with a YO L Ov3 algorithm in a target detection one-stage method;
secondly, detecting a building target in the remote sensing image of the unmanned aerial vehicle by adopting a sliding window and NMS method;
thirdly, inputting the detected building target into a YO L Ov3 deep learning network, and training and optimizing the YO L Ov3 deep learning network.
The invention discloses the following technical effects:
(1) the unmanned aerial vehicle remote sensing image with centimeter-level resolution in the whole area is obtained through the unmanned aerial vehicle, and illegal building data with different characteristics, shapes and types, such as illegal building commercial houses, rural auxiliary houses, industrial enterprise color steel sheds and the like, are drawn on the whole area, so that the illegal building data can be up to millions of illegal data, and the data refinement degree and the data precision can be the first index in the whole country.
(2) The method and the means for deep learning are utilized to extract the buildings against regulations, so that a faster breakthrough is obtained in the aspect of identification efficiency, a more accurate result is obtained in the aspect of identification effect, and the timeliness and the practicability of the extraction of the buildings against regulations are further enhanced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of the violation building intelligent extraction method based on unmanned aerial vehicle remote sensing and deep learning;
FIG. 2 is a schematic view of a sliding window according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of the sliding window detection results before and after being post-processed by the NMS algorithm according to the embodiment of the present invention; wherein, fig. 3(a) is a schematic diagram before performing post-processing on the sliding window detection result by using the NMS algorithm; fig. 3(b) is a schematic diagram after post-processing the sliding window detection result using the NMS algorithm.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Referring to fig. 1 to 3, the embodiment provides an intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning, which includes the following steps:
step S1, collecting remote sensing image data of the unmanned aerial vehicle of the illegal building;
in the remote sensing image data collection stage, the characteristics that the violation building target naturally has multiple scales and irregular boundaries are considered, so that a multiple-scale sample library needs to be constructed, the sample library can cover targets under different scales, meanwhile, the imaging characteristics of the ground object target on a multi-source unmanned aerial vehicle image have natural differences in consideration of the multiple sources (including multiple sensors and multiple time equality) acquired by the unmanned aerial vehicle remote sensing image, therefore, the robustness of a target detection model can be ensured from a data source through the construction of the multi-source multiple-scale violation building target sample library, and in the manufacturing process of a target sample, the remote sensing image from a domestic high-resolution satellite high-grade second GF-2, a resource third-ZY-3 and a WOR L D VIEW3 and the remote sensing image from a Google earth are selected.
Step S2, carrying out remote sensing image preprocessing and sample preparation on the illegal building;
because the original Data format of the remote sensing image of the unmanned aerial vehicle is more, information such as spatial reference is added in the manual marking process conveniently, the output common image Data format is tif, the unmanned aerial vehicle remote sensing image preprocessing model based on a GDA L (geographic space Data Abstraction library) raster Data model is constructed in the embodiment, comprehensive processes such as a coordinate system and affine geographic transformation are fully considered to form a single abstract Data model, integration of reading, writing, converting and processing processes is supported, the collected remote sensing image Data of the violation building is preprocessed through the unified remote sensing image preprocessing model of the unmanned aerial vehicle, and the processing efficiency of the remote sensing image Data of the unmanned aerial vehicle can be effectively improved.
The method comprises the steps of firstly, manually intercepting the preprocessed unmanned aerial vehicle remote sensing image to obtain an image block of a violation building, secondly, labeling the image block by adopting manual marks based on violation commercial houses, rural auxiliary houses, industrial enterprise color steel sheds and other violation building data with different characteristics, different shapes and various types drawn by a manual mode universe, and thirdly, rotationally expanding the image block in multiple directions to solve the problem of diversity, and enabling a deep learning network to fully learn the change of a target while increasing the sample amount, wherein the sample standard refers to a standard format of a public data set VOC (volatile organic compound) for deep learning abroad, the sample is a 256 × 256 thumbnail image, and an XM L file records the basic information of the sample image and the image position of each building target in the sample image.
Step S3, sample data is divided;
and according to the proportion of 7:2:1, dividing the sample library obtained in the step S2 into a training set, a testing set and a cross validation set, so as to prepare data for the training and testing of the subsequent deep learning network model.
S4, constructing a violation building identification model based on the deep learning network;
the embodiment is based on a sliding window and NMS (Non-Maximum Suppression) method, and combines with the YO L Ov3 algorithm in the target detection one-stage method to construct a violation building identification model, thereby obviously reducing the forward propagation time and improving the calculation efficiency of the model when the model is used for target detection.
Firstly, detecting a building target in an unmanned aerial vehicle remote sensing image by adopting a sliding window and NMS method;
compared with a natural image, the data volume of the unmanned aerial vehicle remote sensing image is huge, one unmanned aerial vehicle remote sensing image can reach more than 10000 × 10000 pixels, but a deep learning network model is generally trained based on a small graph (such as 512 × 512), so that sliding window processing must be performed when building target detection is performed, the overlapping degree of two adjacent sliding windows is 50% in the embodiment, wherein a schematic diagram of the sliding windows is shown in fig. 2, and it can be known from fig. 2 that a sliding window 1 is a first sliding window, a sliding window 2 is a second sliding window, and target detection is performed in each sliding window in sequence.
Since the overlapping area of adjacent windows is detected many times as shown in fig. 3(a), it is necessary to perform non-maximum suppression processing on the sliding window detection result using the NMS algorithm, and the processed image is shown in fig. 3 (b).
Secondly, inputting the detected building target into a YO L Ov3 deep learning network, and training the YO L Ov3 deep learning network, wherein the specific training process comprises the following steps:
1) pre-training the top 20 convolutional layers +1 average pooling layer +1 full-link layer of the YO L Ov3 deep learning network were trained using 1000 classes of data in the ImageNet image library and the resolution of the training image data was adjusted to 224 × 224.
2) Training, namely initializing network parameters of the first 20 convolutional layers of the YO L Ov3 deep learning network by using the first 20 convolutional layer network parameters obtained in the step 1), and then training the YO L Ov3 deep learning network by using the training set in the step S3.
3) And optimizing, namely performing cross validation on the trained YO L Ov3 deep learning network by using the cross validation set in the step S3, searching and optimizing the hyper-parameters of the YO L Ov3 network by adopting a random search strategy, performing precision evaluation on the YO L Ov3 deep learning network model obtained by training under different hyper-parameters by using the test set in the step S3, and selecting the YO L Ov3 deep learning network model with the highest precision.
Step S5, extracting the illegal buildings:
inputting the unmanned aerial vehicle remote sensing image to be detected into the unmanned aerial vehicle remote sensing image preprocessing model in the step S2 for preprocessing, and processing the unmanned aerial vehicle remote sensing image to be detected into a standard format capable of inputting a violation building identification model; and inputting the preprocessed data into the illegal building identification model, and identifying, positioning and extracting the illegal building based on the trained illegal building identification model to obtain the illegal building in the unmanned aerial vehicle remote sensing image.
According to the invention, on the obtained unmanned aerial vehicle remote sensing image, a deep neural network specially applied to illegal building identification is trained by manually marking a large number of illegal buildings, and the large-range unknown unmanned aerial vehicle remote sensing image is segmented by an automatic and man-machine interaction method, so that the intelligent identification and positioning of the illegal buildings are realized.
In the description of the present invention, it is to be understood that the terms "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like, indicate orientations or positional relationships based on those shown in the drawings, are merely for convenience of description of the present invention, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and thus, are not to be construed as limiting the present invention.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (6)
1. The intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning is characterized by comprising the following steps:
collecting remote sensing image data of the multi-source multi-scale violation building unmanned aerial vehicle;
constructing an unmanned aerial vehicle remote sensing image preprocessing model, preprocessing collected unmanned aerial vehicle remote sensing image data based on the unmanned aerial vehicle remote sensing image preprocessing model, and making a sample based on the preprocessed unmanned aerial vehicle remote sensing image data to obtain a sample library;
an NMS (network management system) method is suppressed based on a sliding window and a non-maximum value, and a violation building identification model is established by combining a deep learning network; training and optimizing the illegal building identification model by using sample data in the sample library;
inputting the unmanned aerial vehicle remote sensing image to be detected into an unmanned aerial vehicle remote sensing image preprocessing model for preprocessing, inputting the preprocessed data into a trained violation building identification model, and completing identification, positioning and extraction of violation buildings to obtain the violation buildings in the unmanned aerial vehicle remote sensing image to be detected.
2. The method for intelligently extracting the violation buildings based on unmanned aerial vehicle remote sensing and deep learning of claim 1, wherein an unmanned aerial vehicle remote sensing image preprocessing model is constructed based on a geographic space data abstraction library GDA L raster data model, and image data preprocessed by the unmanned aerial vehicle remote sensing image preprocessing model is tif format data.
3. The intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning of claim 1, wherein the sample preparation process comprises the following steps:
firstly, intercepting a preprocessed unmanned aerial vehicle remote sensing image to obtain an image block of a violation building;
secondly, labeling the image block based on the illegal building data sketched in the universe;
and thirdly, performing rotation expansion on the image block in a plurality of directions.
4. The unmanned aerial vehicle remote sensing and deep learning-based intelligent illegal building extraction method according to claim 3, wherein the XM L file of the sample data is used for recording label information of the sample image and the image position of each building target in the sample image.
5. The method for intelligently extracting the illegal building based on unmanned aerial vehicle remote sensing and deep learning of claim 1, wherein the sample library is divided into a training set, a testing set and a cross validation set.
6. The intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning of claim 1, characterized in that the construction and training of the illegal building recognition model specifically comprises the following steps:
firstly, constructing a violation building identification model based on a sliding window and non-maximum value suppression NMS method and combining with a YO L Ov3 algorithm in a target detection one-stage method;
secondly, detecting a building target in the remote sensing image of the unmanned aerial vehicle by adopting a sliding window and NMS method;
thirdly, inputting the detected building target into a YO L Ov3 deep learning network, and training and optimizing the YO L Ov3 deep learning network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010326763.0A CN111507296A (en) | 2020-04-23 | 2020-04-23 | Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010326763.0A CN111507296A (en) | 2020-04-23 | 2020-04-23 | Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111507296A true CN111507296A (en) | 2020-08-07 |
Family
ID=71876332
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010326763.0A Pending CN111507296A (en) | 2020-04-23 | 2020-04-23 | Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111507296A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112016848A (en) * | 2020-09-11 | 2020-12-01 | 范玲珍 | Intelligent detection management system for quality supervision, acceptance and acceptance of constructional engineering based on data scheduling |
| CN112037025A (en) * | 2020-09-01 | 2020-12-04 | 中国银行股份有限公司 | Unmanned aerial vehicle-based bank potential public customer identification method, device and equipment |
| CN112215189A (en) * | 2020-10-21 | 2021-01-12 | 南京智慧航空研究院有限公司 | Accurate detecting system for illegal building |
| CN112215190A (en) * | 2020-10-21 | 2021-01-12 | 南京智慧航空研究院有限公司 | Illegal building detection method based on YOLOV4 model |
| CN112329550A (en) * | 2020-10-16 | 2021-02-05 | 中国科学院空间应用工程与技术中心 | Weak supervision learning-based disaster-stricken building rapid positioning evaluation method and device |
| CN112651338A (en) * | 2020-12-26 | 2021-04-13 | 广东电网有限责任公司电力科学研究院 | Method and device for distinguishing hidden danger of illegal construction of power transmission line |
| CN113011405A (en) * | 2021-05-25 | 2021-06-22 | 南京柠瑛智能科技有限公司 | Method for solving multi-frame overlapping error of ground object target identification of unmanned aerial vehicle |
| CN115797775A (en) * | 2022-12-14 | 2023-03-14 | 中国铁塔股份有限公司重庆市分公司 | Intelligent illegal building identification method and system based on near-earth video image |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110544293A (en) * | 2019-07-15 | 2019-12-06 | 同济大学 | Building scene recognition method based on multi-unmanned aerial vehicle visual cooperation |
| CN110852164A (en) * | 2019-10-10 | 2020-02-28 | 安徽磐众信息科技有限公司 | YOLOv 3-based method and system for automatically detecting illegal building |
-
2020
- 2020-04-23 CN CN202010326763.0A patent/CN111507296A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110544293A (en) * | 2019-07-15 | 2019-12-06 | 同济大学 | Building scene recognition method based on multi-unmanned aerial vehicle visual cooperation |
| CN110852164A (en) * | 2019-10-10 | 2020-02-28 | 安徽磐众信息科技有限公司 | YOLOv 3-based method and system for automatically detecting illegal building |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112037025A (en) * | 2020-09-01 | 2020-12-04 | 中国银行股份有限公司 | Unmanned aerial vehicle-based bank potential public customer identification method, device and equipment |
| CN112037025B (en) * | 2020-09-01 | 2023-08-11 | 中国银行股份有限公司 | Method, device and equipment for identifying potential public customers of bank based on unmanned aerial vehicle |
| CN112016848A (en) * | 2020-09-11 | 2020-12-01 | 范玲珍 | Intelligent detection management system for quality supervision, acceptance and acceptance of constructional engineering based on data scheduling |
| CN112016848B (en) * | 2020-09-11 | 2021-04-06 | 黑龙江省公路工程监理咨询有限公司 | Intelligent detection management system for quality supervision, acceptance and acceptance of constructional engineering based on data scheduling |
| CN112329550A (en) * | 2020-10-16 | 2021-02-05 | 中国科学院空间应用工程与技术中心 | Weak supervision learning-based disaster-stricken building rapid positioning evaluation method and device |
| CN112215189A (en) * | 2020-10-21 | 2021-01-12 | 南京智慧航空研究院有限公司 | Accurate detecting system for illegal building |
| CN112215190A (en) * | 2020-10-21 | 2021-01-12 | 南京智慧航空研究院有限公司 | Illegal building detection method based on YOLOV4 model |
| CN112651338A (en) * | 2020-12-26 | 2021-04-13 | 广东电网有限责任公司电力科学研究院 | Method and device for distinguishing hidden danger of illegal construction of power transmission line |
| CN113011405A (en) * | 2021-05-25 | 2021-06-22 | 南京柠瑛智能科技有限公司 | Method for solving multi-frame overlapping error of ground object target identification of unmanned aerial vehicle |
| CN113011405B (en) * | 2021-05-25 | 2021-08-13 | 南京柠瑛智能科技有限公司 | Method for solving multi-frame overlapping error of ground object target identification of unmanned aerial vehicle |
| CN115797775A (en) * | 2022-12-14 | 2023-03-14 | 中国铁塔股份有限公司重庆市分公司 | Intelligent illegal building identification method and system based on near-earth video image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111507296A (en) | Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning | |
| CN108510467B (en) | SAR image target identification method based on depth deformable convolution neural network | |
| Li et al. | Road network extraction via deep learning and line integral convolution | |
| CN106909902B (en) | Remote sensing target detection method based on improved hierarchical significant model | |
| CN108596108B (en) | Aerial remote sensing image change detection method based on triple semantic relation learning | |
| CN105160310A (en) | 3D (three-dimensional) convolutional neural network based human body behavior recognition method | |
| CN105528595A (en) | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images | |
| CN106844739B (en) | Remote sensing image change information retrieval method based on neural network collaborative training | |
| CN113221625B (en) | Method for re-identifying pedestrians by utilizing local features of deep learning | |
| CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
| CN110532961B (en) | Semantic traffic light detection method based on multi-scale attention mechanism network model | |
| CN107808375B (en) | Merge the rice disease image detecting method of a variety of context deep learning models | |
| CN106408030A (en) | SAR image classification method based on middle lamella semantic attribute and convolution neural network | |
| CN105701467A (en) | Many-people abnormal behavior identification method based on human body shape characteristic | |
| CN102364497A (en) | Image semantic extraction method applied in electronic guidance system | |
| CN112329559A (en) | Method for detecting homestead target based on deep convolutional neural network | |
| CN109034213B (en) | Hyperspectral image classification method and system based on correlation entropy principle | |
| Wang et al. | Extraction of earthquake-induced collapsed buildings using very high-resolution imagery and airborne lidar data | |
| CN109117739A (en) | One kind identifying projection properties extracting method based on neighborhood sample orientation | |
| Fauvel et al. | Detection of hedges in a rural landscape using a local orientation feature: from linear opening to path opening | |
| CN113033386B (en) | High-resolution remote sensing image-based transmission line channel hidden danger identification method and system | |
| CN113808166B (en) | Single-target tracking method based on clustering difference and depth twin convolutional neural network | |
| CN113723558A (en) | Remote sensing image small sample ship detection method based on attention mechanism | |
| CN114037922B (en) | Aerial image segmentation method based on hierarchical context network | |
| CN104504409B (en) | A kind of ancient wall disease identification method based on Global Dictionary feature |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200807 |
|
| RJ01 | Rejection of invention patent application after publication |