CN115861328A - Grave detection method and device and electronic equipment - Google Patents
Grave detection method and device and electronic equipment Download PDFInfo
- Publication number
- CN115861328A CN115861328A CN202310182458.2A CN202310182458A CN115861328A CN 115861328 A CN115861328 A CN 115861328A CN 202310182458 A CN202310182458 A CN 202310182458A CN 115861328 A CN115861328 A CN 115861328A
- Authority
- CN
- China
- Prior art keywords
- grave
- detection
- sample
- burial
- trained
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 531
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000012360 testing method Methods 0.000 claims description 158
- 238000009933 burial Methods 0.000 claims description 118
- 238000011156 evaluation Methods 0.000 claims description 83
- 238000012549 training Methods 0.000 claims description 69
- 238000012795 verification Methods 0.000 claims description 62
- 238000004590 computer program Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 abstract description 3
- 238000013527 convolutional neural network Methods 0.000 description 10
- 238000011835 investigation Methods 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000011160 research Methods 0.000 description 9
- 241000282414 Homo sapiens Species 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000009412 basement excavation Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000002688 persistence Effects 0.000 description 3
- 239000002689 soil Substances 0.000 description 3
- 230000003595 spectral effect Effects 0.000 description 3
- 238000012876 topography Methods 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a grave detection method, a device and electronic equipment, and relates to the field of image processing, wherein the method comprises the following steps: acquiring remote sensing image data of a target area; inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; and determining label information corresponding to the grave sample according to the grave sample and a preset grave information library. The method of the embodiment of the invention effectively improves the detection efficiency and accuracy of the grave.
Description
Technical Field
The invention relates to the field of image processing, in particular to a grave detection method, a grave detection device and electronic equipment.
Background
The trails in China are numerous and rich in form, compared with character materials, the trails can visually show the real historical development conditions of human beings, and the trails are indispensable precious resources and important materials for researching human literature, social science, natural science and the like in China. The grave is a common relic and is also a main research object in the relic detection.
Among the correlation technique, mainly acquire the data in kind through traditional archaeology investigation modes such as field on-the-spot investigation, excavation, nevertheless because the grave is a lot of sparse for distributing, in addition complicated topography and landform and persistence state, adopt traditional archaeology investigation mode length consumption, input greatly, work load is big, and efficiency is lower, can not satisfy the demand of developing large tracts of land grave in the short time and detecting.
Disclosure of Invention
Aiming at the problems in the prior art, the embodiment of the invention provides a grave detection method, a device and electronic equipment.
Specifically, the embodiment of the invention provides the following technical scheme:
in a first aspect, an embodiment of the present invention provides a grave detection method, including:
acquiring remote sensing image data of a target area;
inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking the circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training samples and the grave verification samples are used for training the initial grave detection models to obtain the trained grave detection models; the grave test sample is used for testing each trained grave detection model.
Further, the plurality of initial grave detection models includes at least two of: fast regional proposal network model fast R-CNN, cascade regional proposal network model Cascade R-CNN, yolov5 model and Yolov7 model; the target grave detection model is determined based on the following mode:
respectively inputting a plurality of burial training samples and verification samples into each initial burial detection model, and respectively outputting the detection result of each burial verification sample by each initial burial detection model;
training each initial grave detection model for multiple times, and obtaining each trained grave detection model according to the detection result of each grave verification sample and the label information corresponding to each grave verification sample;
and obtaining the target grave detection model according to the trained grave detection models and the plurality of grave test samples.
Further, the obtaining the target grave detection model according to the trained grave detection models and the plurality of grave test samples includes:
respectively inputting the plurality of burial test samples into the trained burial detection models to obtain the detection results of the burial test samples output by the trained burial detection models;
obtaining an average detection precision evaluation result and a detection rate evaluation result of each trained grave detection model according to the detection result of each grave test sample output by each trained grave detection model and label information corresponding to each grave test sample;
and determining the target grave detection model from the trained grave detection models according to the average detection precision evaluation result and the detection rate evaluation result of each trained grave detection model.
Further, the obtaining of the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model according to the detection result of each grave test sample output by each trained grave detection model and the label information corresponding to each grave test sample comprises:
determining a first quantity, a second quantity and a third quantity according to detection results of the burial test samples output by the trained burial detection models and label information corresponding to the burial test samples; the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third number represents the number of non-graves detected as graves in the detection results of the grave test samples;
determining the average detection accuracy evaluation result of each trained grave detection model according to the first quantity, the second quantity and the third quantity;
and determining the detection rate evaluation result of each trained grave detection model according to the time-consuming duration of the detection result output by each trained grave detection model.
Further, the determining an average detection accuracy evaluation result of each trained grave detection model according to the first number, the second number and the third number includes:
determining the average detection accuracy evaluation result of each trained grave detection model by using the following formula:
wherein AP represents the average detection accuracy evaluation result,
，/>
TP represents said first number, FN represents said second number, FP represents said third number.
In a second aspect, an embodiment of the present invention further provides a grave detection apparatus, including:
the acquisition module is used for acquiring remote sensing image data of a target area;
the detection module is used for inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking the circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test sample is used for testing each trained grave detection model.
In a third aspect, an embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor executes the computer program to implement the burial detection method according to the first aspect.
In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the burial detection method according to the first aspect.
In a fifth aspect, the embodiment of the present invention further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the grave detection method according to the first aspect is implemented.
According to the method, the device and the electronic equipment for detecting the grave, provided by the embodiment of the invention, the remote sensing image data of the target area are obtained, and the remote sensing image data are input into the target grave detection model, so that the grave detection result in the target area can be quickly and accurately obtained. The target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; in the selection process of the grave sample, the circular characteristic which is common in the grave is taken as a main distinguishing mark in the identification and detection process by fully capturing the circular characteristic of the grave, and the circular object in the remote sensing image data is taken as the grave sample, so that the initial grave model can be accurately trained based on the grave sample and the corresponding label information, and the trained grave model can be rapidly and accurately used for detecting the grave; on the other hand, after obtaining the grave sample, divide the grave sample into grave training sample, grave verification sample and grave test sample according to preset proportion, and then train each initial grave detection model based on the grave training sample, the grave verification sample, obtain each grave detection model after training, the grave detection model after training tests based on the grave test sample, regard the grave detection model that the test result is optimal as the target grave detection model, also can obtain more accurate grave detection result, satisfy the extensive detection of grave, promote the detection efficiency and the accuracy of grave. Compared with the traditional field archaeology, the grave detection method in the embodiment of the invention has the characteristics of wide coverage range, high space-time resolution, high spectral resolution and the like, and can obtain a large amount of information which cannot be obtained from ground observation, especially in areas with inconvenient traffic, difficult arrival of human tracks, bad natural conditions and complex landforms. The distribution of the relics is detected by means of the remote sensing image, so that the workload and the cost are greatly reduced, and the detection efficiency and the accuracy of the grave are effectively improved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a burial detection method according to an embodiment of the invention;
FIG. 2 is another schematic flow chart of a burial detection method according to an embodiment of the invention;
fig. 3 is a schematic structural diagram of a grave detection apparatus provided by an embodiment of the invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
The method provided by the embodiment of the invention can be applied to archaeological scenes, and can be used for realizing the rapid and accurate detection of graves.
In the correlation technique, mainly acquire the data in kind through traditional archaeology investigation modes such as open-air on-the-spot investigation, excavation, nevertheless because the more for distributing sparse of tomb quantity, in addition complicated topography and landform and persistence state, adopt traditional archaeology investigation mode long-consuming time, the input is big, work load is big, and efficiency is lower, can not satisfy the demand of developing the large tracts of land tomb detection in the short time.
According to the grave detection method provided by the embodiment of the invention, the remote sensing image data of the target area is obtained, and the remote sensing image data is input into the target grave detection model, so that the grave detection result in the target area is quickly and accurately obtained. The target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; in the selection process of the grave sample, the circular characteristic which is common in the grave is taken as a main distinguishing mark in the identification and detection process by fully capturing the circular characteristic of the grave, and the circular object in the remote sensing image data is taken as the grave sample, so that the initial grave model can be accurately trained based on the grave sample and the corresponding label information, and the trained grave model can be rapidly and accurately used for detecting the grave; on the other hand, after obtaining the grave sample, divide the grave sample into the grave training sample according to the proportion that predetermines, grave verifies sample and grave test sample, and then verify the sample and train each initial grave detection model based on the grave training sample, grave, obtain each grave detection model after the training, the grave detection model after training tests based on the grave test sample, the grave detection model that will test the result the optimum is as the target grave detection model, also can obtain more accurate grave detection result, satisfy the extensive detection of grave, promote the detection efficiency and the accuracy of grave. Compared with the traditional field archaeology, the grave detection method in the embodiment of the invention has the characteristics of wide coverage range, high space-time resolution, high spectral resolution and the like, and can obtain a large amount of information which cannot be obtained from ground observation, especially in areas with inconvenient traffic, difficult arrival of human tracks, bad natural conditions and complex landforms. The distribution of the relics is detected by means of the remote sensing image, so that the workload and the cost are greatly reduced, and the detection efficiency and the accuracy of the grave are effectively improved.
The technical solution of the present invention is described in detail with specific embodiments in conjunction with fig. 1-4. These several specific embodiments may be combined with each other below, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 1 is a schematic flow chart of an embodiment of a burial detection method according to an embodiment of the invention. As shown in fig. 1, the method provided in this embodiment includes:
specifically, in the correlation technique, mainly acquire the material object data through traditional archaeological investigation modes such as open-air on-the-spot investigation, excavation, nevertheless because the burial is more for distributing sparse, and complicated topography and persistence state in addition, adopt traditional archaeological investigation mode to consume long time, invest greatly, work load is big, and efficiency is lower, can not satisfy the demand of developing large tracts of land burial in the short time and detecting.
In order to solve the problem, in the embodiment of the invention, remote sensing image data of a target area is obtained firstly; optionally, the target area is the area to be subjected to the grave detection; alternatively, the remote sensing image data may be Google Earth images.
102, inputting remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test sample is used for testing each trained grave detection model.
Specifically, after the remote sensing image data of the target area are obtained, the remote sensing image data of the target area are input into the target grave detection model in the embodiment of the invention, and a grave detection result in the target area is obtained; the target grave detection model is used for identifying graves in the target area and position information of the graves and displaying the position information in a visualized manner; optionally, the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; alternatively, the plurality of initial grave detection models may be a plurality of types of deep learning models, such as single-stage and two-stage deep learning models; alternatively, the grave sample may be determined as follows: acquiring remote sensing image data of a first area, and taking a circular object in the remote sensing image data of the first area as a burial sample; for example, circular objects such as a circular mound and a circular soil are selected from the Google Earth image, and a circular building can also be used as the grave sample as long as the circular objects are displayed in the Google Earth image, that is, the grave sample can cover different types of circular objects in the Google Earth image. The circular feature of the burial is fully captured, the common feature-circular shape in the burial is used as a main distinguishing mark in the identification and detection process, and the circular object in the remote sensing image data is used as a burial sample.
For example, in the embodiment of the present invention, when determining a grave sample, a circular object is selected from the Google Earth image, that is, a circular feature of the grave is fully captured, a common feature in the grave, namely, a circular shape, is used as a main distinguishing mark in the identification and detection process, and the circular object in the remote sensing image data is used as the grave sample. Optionally, 1640 pieces of grave samples are obtained by using the circular piles, the circular soil and the circular buildings in the remote sensing images, and the size of each piece is 640 × 640. Then, according to the following 8:1: 1. the 1640 grave samples are randomly divided into a training set, a verification set and a test set, wherein 1312 images in the training set are contained, 164 images in the verification set and 164 images in the test set are contained in the training set and the verification set respectively, and the three sets are completely independent, so that a circular grave data set is constructed.
Optionally, after the grave sample is obtained, the grave sample can be compared with grave information in a preset grave information library, so as to determine label information corresponding to the grave sample; optionally, the preset grave information base includes image information and position information of the determined round grave in the first area in the old archaeological process; optionally, the label information is used for marking whether the grave sample is grave; optionally, in a case where the image information and the position information of the grave are compared with each other in the preset grave information library, the label information corresponding to the grave sample is "grave", and in a case where the image information and the position information of the grave sample are not compared with each other in the preset grave information library, that is, in a case where the image information and the position information corresponding to the grave sample cannot be found from the image information and the position information of the circular grave in the preset grave information library, the label information corresponding to the grave sample is "non-grave". Optionally, after the grave sample is obtained, in the embodiment of the invention, the grave sample is divided into a grave training sample, a grave verification sample and a grave test sample according to a preset proportion; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; optionally, the grave training samples are used for training each grave detection model, and the grave verification samples are used for determining the hyper-parameters of each initial grave detection model; the grave test sample is used for testing the trained grave detection model to obtain a test result of the trained grave detection model, so that a target grave detection model is determined; optionally, the grave detection model with the optimal test result in the trained grave detection models is used as the target grave detection model, and then a more accurate grave detection result can be obtained based on the target grave detection model.
According to the method, the remote sensing image data of the target area are obtained, and the remote sensing image data are input into the target burial detection model, so that the burial detection result in the target area can be obtained quickly and accurately. The target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; in the selection process of the grave sample, the circular characteristic which is common in the grave is taken as a main distinguishing mark in the identification and detection process by fully capturing the circular characteristic of the grave, and the circular object in the remote sensing image data is taken as the grave sample, so that the initial grave model can be accurately trained based on the grave sample and the corresponding label information, and the trained grave model can be rapidly and accurately used for detecting the grave; on the other hand, after obtaining the grave sample, divide the grave sample into grave training sample according to the proportion that predetermines, grave verification sample and grave test sample, and then verify the sample and train each initial grave detection model based on the grave training sample, the grave, obtain each grave detection model after training, the grave detection model after training is tested based on the grave test sample, regard the initial grave detection model that the test result is optimal as the target grave detection model, also can obtain more accurate grave detection result, satisfy the extensive detection of grave, promote the detection efficiency and the accuracy of grave. Compared with the traditional field archaeology, the grave detection method in the embodiment of the invention has the characteristics of wide coverage range, high space-time resolution, high spectral resolution and the like, and can obtain a large amount of information which cannot be obtained from ground observation, especially in areas with inconvenient traffic, difficult arrival of human tracks, bad natural conditions and complex landforms. The distribution of the trails is detected by means of the remote sensing image, so that the workload and the cost are greatly reduced, and the detection efficiency and the accuracy of the grave are effectively improved.
In an embodiment, the plurality of initial grave detection models comprises at least two of: a rapid regional proposal network model, a cascade regional proposal network model, a Yolov5 model and a Yolov7 model; the target grave detection model is determined based on the following mode:
respectively inputting a plurality of burial training samples and verification samples into each initial burial detection model, and respectively outputting the detection result of each burial verification sample by each initial burial detection model;
training each initial grave detection model for multiple times, and obtaining each trained grave detection model according to the detection result of each grave verification sample and the label information corresponding to each grave verification sample;
and obtaining a target grave detection model according to the trained grave detection models and the plurality of grave test samples.
Specifically, the plurality of initial grave detection models in the implementation of the invention comprises at least two of the following: the method comprises a fast regional proposal network model (Faster R-CNN), a Cascade regional proposal network model (Cascade R-CNN), a YOLOv5 model and a YOLOv7 model, namely solving the problems of grave classification and prediction in archaeology through a convolutional neural network in the embodiment of the invention, realizing the representation, extraction and learning of grave characteristics, having higher processing speed, obtaining higher precision than the traditional random guessing and support vector machine method, and realizing large-scale accurate detection of graves.
Optionally, in the embodiment of the invention, a plurality of grave training samples and verification samples are respectively input into each initial grave detection model, and each initial grave detection model respectively outputs the detection result of each grave verification sample; optionally, the detection result of each burial verification sample comprises that each burial verification sample is detected to belong to the burial or is detected not to belong to the burial. Optionally, after obtaining the detection results of the burial verification samples respectively output by the initial burial detection models, comparing the detection results of the burial verification samples with the label information corresponding to the burial verification samples, and training the initial burial detection models according to the comparison results to obtain the trained burial detection models; optionally, under the condition that the comparison between the detection result of the grave verification sample and the label information corresponding to the grave verification sample is inconsistent, adjusting and optimizing parameters in the model.
Optionally, under the condition that each initial grave detection model is trained according to the detection result of each grave verification sample and the label information corresponding to each grave verification sample to obtain each trained grave detection model, each trained grave detection model is further tested through a plurality of grave test samples in the embodiment of the invention, so that the target grave detection model is determined; optionally, the grave detection model with the optimal test result is used as the target grave detection model, so that a more accurate grave detection result can be obtained, the large-scale detection of graves is met, and the grave detection efficiency and accuracy are improved.
The method of the embodiment obtains the detection results of the burial verification samples output by the initial burial detection models respectively by inputting the plurality of burial training samples and the verification samples into the initial burial detection models respectively, compares the detection results of the burial verification samples with the label information corresponding to the burial verification samples, trains the initial burial detection models according to the comparison results, and obtains the trained burial detection models; further through a plurality of grave test samples, each grave detection model after training is tested, and the grave detection model with the optimal test result is used as the target grave detection model, so that more accurate grave detection results can be obtained, the large-scale detection of graves is met, and the detection efficiency and accuracy of the graves are improved.
In an embodiment, obtaining the target grave detection model according to the trained grave detection models and the plurality of grave test samples comprises:
respectively inputting the plurality of burial test samples into the trained burial detection models to obtain the detection results of the burial test samples output by the trained burial detection models;
obtaining an average detection precision evaluation result and a detection rate evaluation result of each trained grave detection model according to the detection result of each grave test sample output by each trained grave detection model and the label information corresponding to each grave test sample;
and determining a target grave detection model from the trained grave detection models according to the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model.
Specifically, in the embodiment of the invention, a plurality of burial test samples are respectively input into each trained burial detection model, so as to obtain the detection result of each burial test sample output by each trained burial detection model; optionally, the detection result of each burial test sample comprises that each burial test sample is detected to belong to the burial or is detected not to belong to the burial.
Optionally, after obtaining detection results of the burial test samples respectively output by the trained burial detection models, comparing the detection results of the burial test samples with label information corresponding to the burial test samples to obtain average detection accuracy evaluation results and detection rate evaluation results of the trained burial detection models; optionally, under the condition that the detection result of the grave test sample is consistent with the label information corresponding to each grave test sample, it is indicated that the detection result output by the model is correct; under the condition that the detection result of the grave test sample is inconsistent with the label information corresponding to each grave test sample, indicating that the detection result output by the model is wrong; optionally, the more correct detection results are detected in the test sample, the higher the average detection accuracy evaluation result of the corresponding model is; the more the number of detection result errors in the test sample is, the lower the average detection accuracy evaluation result of the corresponding model is; optionally, the shorter the duration of the output of the detection result of the test sample is, the higher the detection rate evaluation result of the model is; the longer the time length of the detection result output of the test sample is, the lower the detection rate evaluation result of the model is.
Optionally, after determining the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model, determining a target grave detection model from the trained grave detection models according to the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model; optionally, determining a model with the highest average detection accuracy evaluation result and the highest detection rate evaluation result as a target grave detection model; optionally, the average detection accuracy evaluation result of the model may also be considered preferentially, and the model with the highest average detection accuracy evaluation result is selected as the target grave detection model.
According to the method, a plurality of burial test samples are respectively input into the trained burial test models, the detection results of the burial test samples output by the trained burial test models are obtained, then the detection results of the burial test samples are compared with the label information corresponding to the burial test samples, and the average detection accuracy evaluation result and the detection rate evaluation result of the trained burial test models are obtained; further, according to the average detection precision evaluation result and the detection rate evaluation result of each trained grave detection model, the optimal model is selected from the trained grave detection models to serve as the target grave detection model, so that more accurate grave detection results can be obtained, the large-scale detection of graves is met, and the detection efficiency and accuracy of the graves are improved.
In an embodiment, obtaining an average detection accuracy evaluation result and a detection rate evaluation result of each trained grave detection model according to a detection result of each grave test sample output by each trained grave detection model and label information corresponding to each grave test sample, includes:
determining a first quantity, a second quantity and a third quantity according to detection results of the burial test samples output by the trained burial detection models and label information corresponding to the burial test samples; the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-graves detected as graves in the detection results of the grave test samples;
determining the average detection accuracy evaluation result of each trained initial grave detection model according to the first quantity, the second quantity and the third quantity;
and determining the detection rate evaluation result of each trained grave detection model according to the time-consuming duration of each trained grave detection model for outputting the detection result.
Specifically, in order to more reasonably and accurately determine the average detection accuracy evaluation result of each trained grave detection model, the method determines the first number, the second number and the third number according to the detection result of each grave test sample output by each trained grave detection model and the label information corresponding to each grave test sample; wherein the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-graves detected as graves in the detection results of the grave test samples; optionally, the initial values of the first number, the second number and the third number are all 0; optionally, comparing the detection result of each burial test sample with the label information corresponding to each burial test sample, and adding 1 to the first number when the detection result of the test sample is consistent with the label information corresponding to the test sample; for example, if the detection result of the test sample a is "grave" and the label information corresponding to the test sample a is also "grave", the first number is increased by 1; alternatively, in the case where the detection result of the test sample a is "non-grave", and the label information corresponding to the test sample a is "grave", that is, the grave is detected as non-grave in the detection result of the grave test sample, the second number is increased by 1. Optionally, when the detection result of the test sample a is "grave", and the label information corresponding to the test sample a is "non-grave", that is, under the condition that the non-grave is detected as grave in the detection result of the grave test sample, the third quantity is added by 1, so that the accurate classification of the detection result is realized, and further, based on the detection result of the accurate classification, a more accurate and reasonable average detection precision evaluation result can be obtained, so that a more optimal model can be selected as a target grave detection model, and the detection accuracy of the grave is improved.
Optionally, in the embodiment of the invention, in the process of determining the target grave detection model, not only the average detection accuracy evaluation result of the model is considered, but also the detection rate evaluation result of the model is considered; optionally, determining a detection rate evaluation result of each trained grave detection model according to the time-consuming duration of the detection result output by each trained grave detection model; optionally, the longer the time consumed by the model to output the detection result is, the lower the detection rate evaluation result of the model is; the shorter the time consumed by the model for outputting the detection result is, the higher the detection rate evaluation result of the model is; namely, the target grave detection model is determined from two dimensions of efficiency and accuracy, so that a better model can be determined to serve as the target grave detection model, and the detection accuracy and the detection efficiency of the grave are improved.
In the method of the embodiment, in order to obtain more reasonable and accurate average detection accuracy evaluation results of the trained grave detection models, the first number, the second number and the third number are determined according to the detection results of the grave test samples output by the trained grave detection models and label information corresponding to the grave test samples; wherein the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-graves detected as graves in the detection results of the grave test samples; therefore, accurate classification of the detection results is achieved, and more accurate and reasonable average detection accuracy evaluation results can be obtained based on the detection results of the accurate classification. Furthermore, the detection rate evaluation result of the model is combined, and the target grave detection model is determined from two dimensions of efficiency and accuracy, so that a more optimal model can be determined to serve as the target grave detection model, and the detection accuracy and the detection efficiency of the grave are improved.
In an embodiment, determining the average detection accuracy evaluation result of each trained grave detection model according to the first number, the second number and the third number includes:
determining the average detection accuracy evaluation result of each trained grave detection model by using the following formula:
wherein AP represents the average detection accuracy evaluation result,
，/>
TP denotes a first number, FN denotes a second number, and FP denotes a third number.
Specifically, in order to obtain more reasonable and accurate average detection accuracy evaluation results of the trained grave detection models, the first quantity, the second quantity and the third quantity are determined according to the detection results of the grave test samples output by the trained grave detection models and label information corresponding to the grave test samples; wherein the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-graves detected as graves in the detection results of the grave test samples; therefore, accurate classification of the detection results is achieved, and more accurate and reasonable average detection accuracy evaluation results can be obtained based on the detection results of the accurate classification.
Optionally, after determining the first number, the second number and the third number, using a formula
Accurately determining the average detection accuracy evaluation result of each trained grave detection model; wherein AP represents the average detection accuracy evaluation result and is based on the evaluation result>
，/>
TP represents a first number, FN represents a second number, FP represents a third number; the accurate classification of the detection results is realized based on the first quantity, the second quantity and the third quantity, and then based on the detection results of the accurate classification, a more accurate and reasonable average detection accuracy evaluation result can be obtained, so that a more optimal model can be determined to serve as a target grave detection model, and the detection accuracy of the grave is improved.
According to the method of the embodiment, in order to obtain more reasonable and accurate average detection precision evaluation results of the trained grave detection models, the first quantity, the second quantity and the third quantity are determined according to the detection results of the grave test samples output by the trained grave detection models and the label information corresponding to the grave test samples, so that the accurate classification of the detection results is realized, and further, based on the detection results of the accurate classification, more accurate and reasonable average detection precision evaluation results can be obtained, a more optimal model can be determined to serve as a target grave detection model, and the detection accuracy of the grave is improved.
Illustratively, the grave detection method shown in fig. 2 identifies a circular grave based on a deep target learning detection algorithm. The method comprises the following specific steps:
firstly, determining that a research object is circular grave, a research area is an area A, collecting samples from Google Earth images by combining a mining briefing and other archaeological investigation data, and selecting the circular grave from the samples, wherein the circular grave comprises circular piles, circular soil and crop marks, covers different backgrounds, and has 1640 pieces in total and the sizes of 640 multiplied by 640.
Secondly, according to 8:1:1, randomly dividing the sample into a training set, a verification set and a test set, wherein 1312 images are in the training set, 164 images are in the verification set and 164 images are in the test set respectively, and the three sets are completely independent. And labeling the training set and the verification set by using a labelImg tool, wherein the labeling formats are YOLO and VOC, so that a small circular grave data set is constructed.
Thirdly, combining the research progress of the deep learning target detection algorithm, selecting four more classical and mainstream algorithms, namely a fast region proposal network (Faster R-CNN), a Cascade region proposal network (Cascade R-CNN), YOLOv5 and YOLOv7, respectively training the models in the same environment by using a training set and a verification set, and obtaining the network model for detecting the burial after multiple references.
And finally, placing the test set into a model obtained by training for testing, transversely comparing the experimental results of the four algorithms, and introducing an average detection precision evaluation result and a detection rate evaluation result for evaluation, wherein the calculation formula of the relevant indexes of the average detection precision evaluation result is as follows:
wherein, AP represents the evaluation result of average detection precision,
，/>
TP denotes a first number, FN denotes a second number, and FP denotes a third number. Wherein the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-gravels detected as gravels in the detection results of the grave test samples, so that the detection results are accurately classified, more accurate and reasonable average detection precision evaluation results can be obtained based on the accurately classified detection results, a more optimal model can be determined to serve as a target grave detection model, and the detection accuracy of the gravels is improved.
The embodiment of the invention is based on a deep learning target detection algorithm, effectively utilizes the strong feature expression capability of the deep learning algorithm to automatically identify the circular grave, and realizes the automatic detection of the circular grave on the self-built remote sensing data set. Optionally, in the embodiment of the invention, a remote sensing data set taking circular graves as targets is constructed by fully capturing circular features of the graves, and several models based on CNN are respectively selected from single-stage and two-stage target detection algorithms for experiments, that is, circular and grave positions in remote sensing images are identified based on neural network automatic learning features and inference target area candidate frames. And finally, evaluating the detection performance of different algorithms, and analyzing the difference of detection results, thereby determining a target grave detection model and improving the detection accuracy of the grave. The grave detection method in the embodiment of the invention combines remote sensing archaeology and target detection research, self-establishes a circular grave data set, automatically detects the circular grave by means of high-resolution satellite data, and can be used as an important supplement for field archaeology and the traditional remote sensing archaeology technology.
In addition, the research area and the research object selected in the embodiment of the invention are common and typical in archaeological research, and have universal research significance; in addition, the most common circular features are taken as the main basis in the process of screening and identifying the sample in consideration of the rich form and system of the grave, and the method has strong pertinence. And finally, the detection performance of various target detection algorithms is analyzed, so that the detection accuracy and the detection efficiency of the grave can be effectively improved.
The grave detection apparatus according to the present invention will be described below, and the grave detection apparatus described below and the grave detection method described above can be referred to in correspondence with each other.
Fig. 3 is a schematic structural view of the grave detection apparatus according to the present invention. The grave detection device that this embodiment provided includes:
an obtaining module 710, configured to obtain remote sensing image data of a target area;
the detection module 720 is used for inputting the remote sensing image data into the target grave detection model to obtain a grave detection result in the target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the burial sample includes: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each grave detection model to obtain each trained grave detection model; the grave test sample is used for testing the trained grave detection model.
Optionally, the plurality of initial grave detection models comprises at least two of: a fast regional proposal network model (Faster R-CNN), a Cascade regional proposal network model (Cascade R-CNN), a Yolov5 model and a Yolov7 model;
optionally, the grave detection apparatus provided by the invention further comprises a model training module, which is specifically configured to:
respectively inputting a plurality of burial training samples and verification samples into each initial burial detection model, and respectively outputting the detection result of each burial verification sample by each initial burial detection model;
training each initial grave detection model for multiple times, and obtaining each trained grave detection model according to the detection result of each grave verification sample and the label information corresponding to each grave verification sample;
and obtaining a target grave detection model according to the trained grave detection models and the plurality of grave test samples.
Optionally, the model training module is specifically configured to: respectively inputting the plurality of burial test samples into the trained burial detection models to obtain the detection results of the burial test samples output by the trained burial detection models;
obtaining an average detection precision evaluation result and a detection rate evaluation result of each trained grave detection model according to the detection result of each grave test sample output by each trained grave detection model and label information corresponding to each grave test sample;
and determining a target grave detection model from the trained grave detection models according to the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model.
Optionally, the model training module is specifically configured to: determining a first quantity, a second quantity and a third quantity according to the detection result of each burial test sample output by each trained burial detection model and the label information corresponding to each burial test sample; the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third quantity represents the quantity of non-graves detected as graves in the detection results of the grave test samples;
determining the average detection accuracy evaluation result of each trained grave detection model according to the first quantity, the second quantity and the third quantity;
and determining the detection rate evaluation result of each trained grave detection model according to the time-consuming duration of each trained grave detection model for outputting the detection result.
Optionally, the model training module determines an average detection accuracy evaluation result of each trained initial grave detection model by using the following formula:
wherein AP represents the average detection accuracy evaluation result,
，/>
TP denotes a first number, FN denotes a second number, and FP denotes a third number.
The apparatus of the embodiment of the present invention is configured to perform the method of any of the foregoing method embodiments, and the implementation principle and the technical effect are similar, which are not described herein again.
Fig. 4 illustrates a physical structure diagram of an electronic device, which may include: a processor (processor) 810, a communication Interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication Interface 820 and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call logic instructions in the memory 830 to perform a burial detection method, the method comprising: acquiring remote sensing image data of a target area; inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the burial sample includes: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each grave detection model to obtain each trained grave detection model; the grave test sample is used for testing each trained grave detection model.
In addition, the logic instructions in the memory 830 may be implemented in software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, which when executed by a computer, enable the computer to perform the burial detection method provided by the above methods, the method comprising: acquiring remote sensing image data of a target area; inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the burial sample includes: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test sample is used for testing the trained grave detection model.
In still another aspect, the present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program, which when executed by a processor, is implemented to perform the grave detection methods provided above, the method comprising: acquiring remote sensing image data of a target area; inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking a circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test sample is used for testing the trained grave detection model.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment may be implemented by software plus a necessary general hardware platform, and may also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the various embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (8)
1. A grave detection method is characterized by comprising the following steps:
acquiring remote sensing image data of a target area;
inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking the circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test samples are used for testing the trained grave detection models.
2. A burial detection method as recited in claim 1, wherein the plurality of initial burial detection models includes at least two of: fast regional proposal network model fast R-CNN, cascade regional proposal network model Cascade R-CNN, yolov5 model and Yolov7 model; the target grave detection model is determined based on the following mode:
respectively inputting a plurality of burial training samples and verification samples into each initial burial detection model, and respectively outputting the detection result of each burial verification sample by each initial burial detection model;
training each initial grave detection model for multiple times, and obtaining each trained grave detection model according to the detection result of each grave verification sample and the label information corresponding to each grave verification sample;
and obtaining the target grave detection model according to the trained grave detection models and the plurality of grave test samples.
3. A burial detection method as claimed in claim 2, wherein the obtaining of the target burial detection model based on the trained burial detection models and the plurality of burial test samples, comprises:
respectively inputting the plurality of burial test samples into the trained burial detection models to obtain the detection results of the burial test samples output by the trained burial detection models;
obtaining an average detection precision evaluation result and a detection rate evaluation result of each trained grave detection model according to the detection result of each grave test sample output by each trained grave detection model and label information corresponding to each grave test sample;
and determining the target grave detection model from the trained grave detection models according to the average detection accuracy evaluation result and the detection rate evaluation result of each trained grave detection model.
4. A burial detection method as claimed in claim 3, wherein the obtaining of the average detection accuracy evaluation result and the detection rate evaluation result of each trained burial test model based on the detection result of each burial test sample output by each trained burial test model and the label information corresponding to each of the burial test samples comprises:
determining a first quantity, a second quantity and a third quantity according to the detection result of each burial test sample output by each trained burial detection model and the label information corresponding to each burial test sample; the first number represents the number of graves detected as graves in the detection results of the grave test samples; the second number represents the number of graves detected as non-graves in the detection results of the grave test samples; the third number represents the number of non-graves detected as graves in the detection results of the grave test samples;
determining the average detection accuracy evaluation result of each trained grave detection model according to the first quantity, the second quantity and the third quantity;
and determining the detection rate evaluation result of each trained grave detection model according to the time-consuming duration of each trained grave detection model for outputting the detection result.
5. A burial detection method as claimed in claim 4, wherein the determining of the average detection accuracy evaluation result of each trained burial detection model according to the first number, the second number and the third number comprises:
determining the average detection accuracy evaluation result of each trained grave detection model by using the following formula:
wherein AP represents the average detection accuracy evaluation result,
，/>
TP represents the first number, FN represents the second number, FP represents the third number.
6. A grave detection apparatus, comprising:
the acquisition module is used for acquiring remote sensing image data of a target area;
the detection module is used for inputting the remote sensing image data into a target grave detection model to obtain a grave detection result in a target area; the target grave detection model is determined based on the plurality of initial grave detection models, the grave samples and label information corresponding to the grave samples; the grave sample and the label information are determined based on the following manner: acquiring remote sensing image data of a first area; taking the circular object in the remote sensing image data of the first area as a grave sample; determining label information corresponding to the grave sample according to the grave sample and a preset grave information library; the grave sample comprises: the method comprises the following steps of (1) carrying out a grave training sample, a grave verification sample and a grave test sample; the grave training sample and the grave verification sample are used for training each initial grave detection model to obtain each trained grave detection model; the grave test sample is used for testing each trained grave detection model.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the burial detection method according to any one of claims 1 to 5 when executing the program.
8. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the burial detection method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310182458.2A CN115861328A (en) | 2023-03-01 | 2023-03-01 | Grave detection method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310182458.2A CN115861328A (en) | 2023-03-01 | 2023-03-01 | Grave detection method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115861328A true CN115861328A (en) | 2023-03-28 |
Family
ID=85659456
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310182458.2A Pending CN115861328A (en) | 2023-03-01 | 2023-03-01 | Grave detection method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115861328A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1560635A (en) * | 2004-03-03 | 2005-01-05 | 中国煤炭地质总局航测遥感局 | Remote sensing archaeological detecting method for emperor tomb and paleoichnology |
| JP2020071615A (en) * | 2018-10-30 | 2020-05-07 | 富士通株式会社 | Detection method, detection program, and detection device |
| CN111582004A (en) * | 2019-02-15 | 2020-08-25 | 阿里巴巴集团控股有限公司 | Target area segmentation method and device in ground image |
| CN112733800A (en) * | 2021-01-22 | 2021-04-30 | 中国科学院地理科学与资源研究所 | Remote sensing image road information extraction method and device based on convolutional neural network |
| CN113033315A (en) * | 2021-02-26 | 2021-06-25 | 江西理工大学 | Rare earth mining high-resolution image identification and positioning method |
-
2023
- 2023-03-01 CN CN202310182458.2A patent/CN115861328A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1560635A (en) * | 2004-03-03 | 2005-01-05 | 中国煤炭地质总局航测遥感局 | Remote sensing archaeological detecting method for emperor tomb and paleoichnology |
| JP2020071615A (en) * | 2018-10-30 | 2020-05-07 | 富士通株式会社 | Detection method, detection program, and detection device |
| CN111582004A (en) * | 2019-02-15 | 2020-08-25 | 阿里巴巴集团控股有限公司 | Target area segmentation method and device in ground image |
| CN112733800A (en) * | 2021-01-22 | 2021-04-30 | 中国科学院地理科学与资源研究所 | Remote sensing image road information extraction method and device based on convolutional neural network |
| CN113033315A (en) * | 2021-02-26 | 2021-06-25 | 江西理工大学 | Rare earth mining high-resolution image identification and positioning method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Malambo et al. | Automated detection and measurement of individual sorghum panicles using density-based clustering of terrestrial lidar data | |
| Urbach et al. | Automatic detection of sub-km craters in high resolution planetary images | |
| CN112990086B (en) | Remote sensing image building detection method and device and computer readable storage medium | |
| CN110910343A (en) | Method and device for detecting pavement cracks and computer equipment | |
| Saxena et al. | Capturing heterogeneous urban growth using SLEUTH model | |
| CN110472597A (en) | Rock image rate of decay detection method and system based on deep learning | |
| CN115409069A (en) | Village and town building identification method, classification method, device, electronic equipment and medium | |
| CN111191696B (en) | Deep learning algorithm-based steel bar layering method and system | |
| CN104463240B (en) | A kind of instrument localization method and device | |
| CN107392252A (en) | Computer deep learning characteristics of image and the method for quantifying perceptibility | |
| CN106845559A (en) | Take the ground mulching verification method and system of POI data special heterogeneity into account | |
| CN107563366A (en) | A kind of localization method and device, electronic equipment | |
| CN112990085A (en) | Method and device for detecting change of culture pond and computer readable storage medium | |
| CN114782845B (en) | Method, device, equipment and medium for identifying newly-born cracks of building after blasting | |
| CN117893383B (en) | Urban functional area identification method, system, terminal equipment and medium | |
| CN117095216B (en) | Model training method, system, equipment and medium based on countermeasure generation network | |
| Weidner et al. | An algorithm for measuring landslide deformation in terrestrial lidar point clouds using trees | |
| CN114005027A (en) | Urban single tree detection system and method based on unmanned aerial vehicle image | |
| CN117171533A (en) | Real-time acquisition and processing method and system for geographical mapping operation data | |
| CN117557901A (en) | Detection model of small target crops in field and construction method | |
| Sun et al. | QuickQuakeBuildings: Post-earthquake SAR-Optical Dataset for Quick Damaged-building Detection | |
| KR102416714B1 (en) | System and method for city-scale tree mapping using 3-channel images and multiple deep learning | |
| CN115861328A (en) | Grave detection method and device and electronic equipment | |
| CN114239250B (en) | System and method for territorial space planning design | |
| Zhang et al. | Machine learning methods for Unesco Chinese heritage: Complexity and comparisons |
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: 20230328 |
|
| RJ01 | Rejection of invention patent application after publication |