CN116188502B - Method for dividing infrared image of photovoltaic panel, storage medium and electronic device - Google Patents
Method for dividing infrared image of photovoltaic panel, storage medium and electronic device Download PDFInfo
- Publication number
- CN116188502B CN116188502B CN202310469313.0A CN202310469313A CN116188502B CN 116188502 B CN116188502 B CN 116188502B CN 202310469313 A CN202310469313 A CN 202310469313A CN 116188502 B CN116188502 B CN 116188502B
- Authority
- CN
- China
- Prior art keywords
- photovoltaic panel
- image
- infrared image
- photovoltaic
- segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/50—Photovoltaic [PV] energy
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a photovoltaic panel infrared image segmentation method, a storage medium and electronic equipment, wherein the method comprises the following steps: acquiring a photovoltaic panel infrared image shot by an unmanned aerial vehicle; dividing a photovoltaic panel in the infrared image of the photovoltaic panel to obtain a photovoltaic panel area image and first coordinate information of the photovoltaic panel area in the infrared image of the photovoltaic panel; inputting a photovoltaic panel region image into a pre-trained example segmentation model based on yolov5 to obtain a photovoltaic module region and second coordinate information of the photovoltaic module region in the photovoltaic panel region image, wherein the example segmentation model is provided with at least four groups of anchor parameters, and at least one small target detection layer is added; performing edge coloring on the photovoltaic module area, and performing edge segmentation based on an edge coloring result; and obtaining a photovoltaic module segmentation result of the infrared image of the photovoltaic panel according to the edge segmentation result, the first coordinate information and the second coordinate information. The method can solve various problems of rotation, pitch angle and the like of the image caused by the gesture of the unmanned aerial vehicle, and has higher segmentation precision.
Description
Technical Field
The invention relates to the technical field of image segmentation, in particular to a photovoltaic panel infrared image segmentation method, a storage medium and electronic equipment.
Background
With the rapid development of the photovoltaic industry, a large number of Photovoltaic (PV) solar power plants are facing operational and maintenance challenges. The photovoltaic module has the characteristics of wide distribution, large quantity and high failure rate, and the defects are difficult to clean and position. Thousands of solar panels are usually located in the rare corners of mountain areas and the like, and mainly perform patrol tasks according to specified routes, shifts and specified projects by patrol workers, so that the cost is high and the efficiency is low. Since defects typically manifest as localized overheating or localized temperature anomalies, infrared (IR) imaging is widely used to identify damaged solar panels on the surface temperature of photovoltaic modules, also known as thermal imaging. Thermal imaging may enhance the identification of defects and attenuate interference in background areas compared to visible images. Meanwhile, along with the wide application of unmanned aerial vehicles, more and more photovoltaic power stations begin to attempt to acquire photovoltaic module images by adopting unmanned aerial vehicles carrying infrared cameras and autonomously identify defects by utilizing an image identification technology.
In the related art, roof infrared photovoltaic inspection is carried out, because the color characteristics in the infrared image are single, the characteristics of the infrared photovoltaic module are not obvious, so that the segmentation precision is low, the photovoltaic panel in the infrared image can be identified and detected, and the smaller photovoltaic module in the photovoltaic panel can not be segmented. Meanwhile, because of the flying angle and the gesture of the unmanned aerial vehicle, the rotating inclined photovoltaic panel is caused, and accurate segmentation cannot be performed.
Disclosure of Invention
The present invention aims to solve at least one of the technical problems in the related art to some extent. Therefore, the invention aims to provide a photovoltaic panel infrared image segmentation method, a storage medium and electronic equipment, so as to segment a photovoltaic module from a photovoltaic panel infrared image with high precision, and solve the problem that the image cannot be segmented due to the posture of an unmanned aerial vehicle.
In order to achieve the above object, an embodiment of a first aspect of the present invention provides a method for segmenting an infrared image of a photovoltaic panel, the method comprising: acquiring a photovoltaic panel infrared image shot by an unmanned aerial vehicle; dividing a photovoltaic panel in the infrared image of the photovoltaic panel to obtain a photovoltaic panel area image and first coordinate information of the photovoltaic panel area in the infrared image of the photovoltaic panel; inputting the photovoltaic panel region image into a pre-trained example segmentation model based on yolov5 to obtain a photovoltaic module region and second coordinate information of the photovoltaic module region in the photovoltaic panel region image, wherein the example segmentation model based on yolov5 is provided with at least four groups of anchor parameters, and at least one small target detection layer is added; performing edge coloring on the photovoltaic module area, and performing edge segmentation based on an edge coloring result; and obtaining a photovoltaic module segmentation result of the infrared image of the photovoltaic panel according to the edge segmentation result, the first coordinate information and the second coordinate information.
In addition, the method for segmenting the infrared image of the photovoltaic panel according to the embodiment of the invention can also have the following additional technical characteristics:
according to one embodiment of the invention, the method further comprises: generating a background image, wherein the size of the background image is the same as that of the infrared image of the photovoltaic panel, and the color is a first preset color; and generating a mask image of the photovoltaic module in the infrared image of the photovoltaic panel according to the first coordinate information, the second coordinate information and the background image.
According to an embodiment of the present invention, the edge segmentation is performed on the photovoltaic panel in the infrared image of the photovoltaic panel to obtain a photovoltaic panel area image and first coordinate information of the photovoltaic panel area in the infrared image of the photovoltaic panel, including: performing convolution operation on the infrared image of the photovoltaic panel to position the photovoltaic panel in the infrared image of the photovoltaic panel; fitting the positioning result based on the color characteristics and the shape characteristics of the photovoltaic panel to optimize the positioning result; and obtaining the photovoltaic panel area image and the first coordinate information of the photovoltaic panel area in the photovoltaic panel infrared image according to the optimization result.
According to one embodiment of the present invention, the example segmentation model training process based on yolov5 includes: acquiring a training sample set, wherein the training sample set comprises a positive sample and a negative sample, the positive sample is an infrared image containing a photovoltaic panel, and the negative sample is an infrared image not containing the photovoltaic panel; dividing the photovoltaic panel in the positive sample to obtain a photovoltaic panel image, and marking the photovoltaic module in the photovoltaic panel image; dividing the negative sample and the marked photovoltaic panel image into a training set and a testing set; and constructing the example segmentation model based on the yolov5, training the example segmentation model based on the yolov5 by utilizing the training set, and testing the trained example segmentation model based on the yolov5 by utilizing the testing set to obtain the trained example segmentation model based on the yolov 5.
According to one embodiment of the invention, in training the yolov 5-based instance segmentation model with the test set, the trained yolov 5-based instance segmentation model is tested by lowering a category confidence threshold and a IoU threshold multiple times.
According to one embodiment of the present invention, the edge coloring the photovoltaic module area and the edge segmentation based on the edge coloring result includes: according to the second coordinate information, a second preset color is formed on the edge of the photovoltaic module area in the photovoltaic panel area image; performing color detection on the colored photovoltaic panel area image to extract the outline of the photovoltaic module; and carrying out contour recognition on the photovoltaic module, and carrying out edge segmentation according to the recognized contour.
According to an embodiment of the present invention, the obtaining the photovoltaic module segmentation result of the photovoltaic panel infrared image according to the edge segmentation result, the first coordinate information and the second coordinate information includes: performing expansion and/or corrosion operation on the edge segmentation result, and updating the second coordinate information according to the operation result; and mapping the updated second coordinate information to the infrared image of the photovoltaic panel according to the first coordinate information to obtain a photovoltaic module segmentation result of the infrared image of the photovoltaic panel.
According to one embodiment of the invention, the yolov 5-based example segmentation model is provided with four sets of anchor parameters, respectively [5,6;8,14;15,11] #4, [10,13;16,30;33,23] # P3/8, [30,61;62,45;59,119] # P4/16, [116,90;156,198;373,326] # P5/32.
To achieve the above object, a second aspect of the present invention provides a computer readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned method for segmenting an infrared image of a photovoltaic panel.
To achieve the above object, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory, where the computer program, when executed by the processor, implements the above-mentioned method for segmenting an infrared image of a photovoltaic panel.
The photovoltaic panel infrared image segmentation method, the storage medium and the electronic equipment can solve various problems of rotation, pitch angle and the like of images caused by the gesture of an unmanned aerial vehicle, and the segmentation precision is high.
Drawings
Fig. 1 is a flowchart of a method for dividing an infrared image of a photovoltaic panel according to a first embodiment of the present invention;
FIG. 2 is an infrared image of a photovoltaic panel according to one embodiment of the present invention;
FIG. 3 is a flow chart of a method for segmenting an infrared image of a photovoltaic panel according to a second embodiment of the present invention;
FIG. 4 is a schematic diagram of a single channel single convolution kernel of one embodiment of the present disclosure;
FIG. 5 is a schematic diagram of a convolution operation of one embodiment of the present invention;
FIG. 6 is a photovoltaic panel area image of one embodiment of the present invention;
FIG. 7 is a flow chart of a method for segmenting an infrared image of a photovoltaic panel according to a third embodiment of the present invention;
fig. 8 is a flowchart of a method for dividing an infrared image of a photovoltaic panel according to a fourth embodiment of the present invention;
FIG. 9 is a flow chart of a method for segmenting an infrared image of a photovoltaic panel according to a fifth embodiment of the present invention;
FIG. 10 is a graph of the photovoltaic module segmentation result according to one embodiment of the present invention;
FIG. 11 is a flowchart of a method for infrared image segmentation of a photovoltaic panel according to a sixth embodiment of the present invention;
FIG. 12 is a mask diagram of a photovoltaic module in a photovoltaic panel infrared image according to one embodiment of the present invention;
fig. 13 is a block diagram of an electronic device according to an embodiment of the invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.
The following describes a photovoltaic panel infrared image segmentation method, a storage medium, and an electronic apparatus according to an embodiment of the present invention with reference to the accompanying drawings.
Fig. 1 is a flowchart of a method for dividing an infrared image of a photovoltaic panel according to an embodiment of the present invention.
As shown in fig. 1, the method includes:
s11, acquiring an infrared image of the photovoltaic panel shot by the unmanned aerial vehicle.
Specifically, as shown in fig. 2, the infrared image of the photovoltaic panel may be an image obtained by photographing a photovoltaic panel disposed on a roof, and the resolution of the infrared image of the photovoltaic panel may be 1280×1024.
And S12, dividing the photovoltaic panel in the infrared image of the photovoltaic panel to obtain a photovoltaic panel area image and first coordinate information of the photovoltaic panel area in the infrared image of the photovoltaic panel.
Specifically, as the image recognition or the image segmentation simply cannot meet the requirement, the image recognition and segmentation technology based on yolov5 is adopted in the invention. In the first stage, based on the characteristic information consideration of the whole image (the resolution can be 1280 x 1024), the photovoltaic panel is firstly segmented (in the whole infrared image, the color characteristics of the photovoltaic panel are obviously different from the colors of other areas, and the shape characteristics of the photovoltaic panel are in regular rectangular or polygonal shapes, so that the photovoltaic panel can be positioned according to the two obvious characteristics when convolution operation is carried out, and then edge segmentation is carried out), so that a photovoltaic panel area image with the resolution of 640 x 640 is obtained. Because the proportion of the photovoltaic panel on the original image is larger and the characteristics are more obvious, the influence of some lost characteristics on the whole accuracy is small (can be ignored) when convolution characteristic extraction is carried out, and the time cost is also reduced.
It should be noted that, due to the problems of the flight attitude and the camera angle of the unmanned aerial vehicle, the photographed image may not be positive, and a certain inclination angle may be generated, which results in that the conventional image recognition cannot be applied here. Because the traditional detection frame for target detection is a positive rectangular frame, and the photovoltaic panel in the figure has rotation angle, the positive rectangular detection frame can not be divided along the edge of the component, the requirements are not met, the rotation angle is large, the two detection frames are overlapped, repeated detection is carried out for many times for a small photovoltaic component, and the time cost is increased. For this purpose, the invention employs a segmentation technique in the first stage.
S13, inputting the photovoltaic panel region image into a pre-trained example segmentation model based on yolov5 to obtain a photovoltaic module region and second coordinate information of the photovoltaic module region in the photovoltaic panel region image, wherein the example segmentation model based on yolov5 is provided with at least four groups of anchor parameters, and at least one small target detection layer is added.
The pre-trained example segmentation model based on yolov5 may have an input image resolution of 640 x 640.
And S14, carrying out edge coloring on the photovoltaic module area, and carrying out edge segmentation based on an edge coloring result.
Specifically, after the photovoltaic panel area image is input into the pre-trained yolov 5-based example segmentation model, the output image can detect each photovoltaic module, but the detection frame may not be along the edges of the photovoltaic modules. When the rotation angle is large, a significant error occurs, so that the segmentation is inaccurate. Therefore, the invention performs edge coloring on the photovoltaic module area and performs edge segmentation based on the edge coloring result.
And S15, obtaining a photovoltaic module segmentation result of the infrared image of the photovoltaic panel according to the edge segmentation result, the first coordinate information and the second coordinate information.
According to the method for segmenting the infrared image of the photovoltaic panel, disclosed by the embodiment of the invention, the photovoltaic infrared image is segmented by adopting multiple stages (the photovoltaic panel is segmented in the first stage and the photovoltaic assembly is segmented in the second stage), compared with the traditional segmentation technology, the precision is obviously improved, and meanwhile, the situations of misjudgment, missed detection and the like are avoided.
Specifically, the size of the infrared image of the photovoltaic panel photographed by the unmanned aerial vehicle is generally 1280 x 1024, and the common detection model is 640 x 640, so compression of the image is necessarily caused, the image loses some key point information, so that detection and segmentation accuracy is affected, but if a model structure of 1280 x 1280 is adopted, although loss caused by image compression can be avoided to a certain extent, the processing speed is affected, and the segmentation of the infrared image belongs to segmentation of a small object of multiple objects, the characteristic information of the detection object is still lost after the multilayer rolling and pooling of the traditional network structure, so that the detection segmentation accuracy is affected, and misdetection and omission occur, so that the error is unavoidable. Therefore, in order to solve such a problem, the present embodiment employs a multi-stage roof photovoltaic separation technique. On the premise of shortening the processing time of the model as much as possible, the detection segmentation precision is improved by two times or even three times. Meanwhile, the problem that an image cannot be segmented due to the posture of the unmanned aerial vehicle can be solved.
In some embodiments, as shown in fig. 3, edge-dividing a photovoltaic panel in an infrared image of the photovoltaic panel to obtain an image of a photovoltaic panel region and first coordinate information of the photovoltaic panel region in the infrared image of the photovoltaic panel, including:
s31, carrying out convolution operation on the infrared image of the photovoltaic panel to realize the positioning of the photovoltaic panel in the infrared image of the photovoltaic panel.
Specifically, an image is composed of a plurality of pixels, and a single pixel may be represented by three values of RGB, such as a black RGB value of (0, 0), white (255, 255, 255), and the like. When convolution operation is carried out, the photovoltaic panel and the surrounding colors in the infrared image have obvious mutation on RGB values, and the position of the photovoltaic panel is easy to position; taking a single-channel single convolution kernel as an example, as shown in fig. 4, the input is input, the convolution kernel of 3*3 is kernel, and bias is the offset when the convolution is performed. As shown in fig. 5, each time a convolution operation is performed, the convolution kernel translates bias units to the right until the transverse translation is completed, and translates bias downward again to repeat the previous operation until the convolution of the whole input picture is completed.
And S32, fitting the positioning result based on the color characteristics and the shape characteristics of the photovoltaic panel, so as to optimize the positioning result.
Specifically, fitting is carried out according to the shape and color characteristics of the photovoltaic panel, and objects with darker colors but not the photovoltaic panel in the positioned photovoltaic panel image are removed.
And S33, obtaining a photovoltaic panel area image and first coordinate information of the photovoltaic panel area in the photovoltaic panel infrared image according to the optimization result.
Specifically, as shown in fig. 6, because of the problems of the flying gesture and the camera angle of the unmanned aerial vehicle, the shot picture has a certain inclination angle, the detection frame for target detection is a regular rectangular frame in general, the photovoltaic panel in the figure has a rotation angle, the detection frame of the regular rectangular frame cannot be divided along the edge of the component, the rotation angle of the detection frame is large, the two detection frames can be intersected, the small photovoltaic component is repeatedly detected for a plurality of times, and the time cost is increased, so that the edge division algorithm can be adopted to divide the photovoltaic panel region image from the infrared image of the photovoltaic panel.
In some embodiments, as shown in fig. 7, the example segmentation model based on yolov5 training process includes:
s51, acquiring a training sample set, wherein the training sample set comprises a positive sample and a negative sample, the positive sample is an infrared image containing a photovoltaic panel, and the negative sample is an infrared image not containing the photovoltaic panel.
The positive samples in the training sample set can be obtained through the segmentation step of the first stage.
S52, dividing the photovoltaic panel in the positive sample to obtain a photovoltaic panel image, and marking the photovoltaic module in the photovoltaic panel image.
Specifically, the photovoltaic panel in the positive sample is subjected to segmentation marking. Unlike the traditional marking method (the traditional marking method is to mark the middle part of two photovoltaic modules), the method for directly marking the photovoltaic modules is adopted because the characteristics of the middle part of the two photovoltaic modules in the infrared image are not obvious and are discontinuous, and the generated mask image (mask) can be in a discontinuous condition.
S53, dividing the negative sample and the marked photovoltaic panel image into a training set and a testing set.
S54, constructing an example segmentation model based on yolov5, training the example segmentation model based on yolov5 by using a training set, and testing the trained example segmentation model based on yolov5 by using a testing set to obtain a trained example segmentation model based on yolov 5.
Specifically, the photovoltaic module occupies fewer pixels, and belongs to dense detection of small targets. In order to prevent false detection and missing detection. According to the invention, the network structure of the segmentation model is optimized, and a small target detection layer is added in the example segmentation model based on yolov5, so that the convergence speed of the model is ensured, and the omission ratio is reduced. It should be noted that, one reason why the detection effect of the yolov5 small target is not good is that the small target sample is smaller in size, the downsampling multiple of the yolov5 is larger, and the deeper feature map is difficult to learn the feature information of the small target, so that the invention proposes to increase the small target detection layer to detect after splicing the shallower feature map and the deep feature map.
Based on the example segmentation model of yolov5, the anchor parameter is the size of the anchor frame, and the anchor frame is obtained by setting different aspect ratios according to the size of the component to be detected. In some embodiments, the yolov 5-based example segmentation model is provided with four sets of anchor parameters, respectively [5,6;8,14;15,11] #4, [10,13;16,30;33,23] # P3/8, [30,61;62,45;59,119] # P4/16, [116,90;156,198;373,326] # P5/32. The anchor parameters of the large feature map, the medium feature map and the small feature map are sequentially from front to back, [5,6;8,14;15,11] #4 is the anchor parameter corresponding to the small target detection layer, taking 640 x 640 as an example of input image resolution, #4 represents 160 x 160 as output feature map resolution, namely 640/4, and the large feature map contains more low-level information and is suitable for detecting the small target, and the anchor scale is smaller.
In some embodiments, during training of the yolov 5-based instance segmentation model with the test set, the trained yolov 5-based instance segmentation model is tested by lowering the class confidence threshold and the IoU (Intersection over Union, cross-over) threshold multiple times.
Specifically, the example segmentation model based on yolov5 is tested by using a test set, and the class confidence threshold and the IoU threshold are continuously reduced by confusing the omission ratio of the matrix statistical model to reduce the omission ratio.
The confusion matrix is a summary of prediction results of the classification problem, uses a numerical value to summarize the number of correct predictions and incorrect predictions, and subdivides each class, and shows which part is confused when the classification model predicts. It is convenient to see through this matrix if two different classes are confused (one class is mistaken for the other). The confusion matrix can intuitively understand errors made by the classification model, more importantly, can understand which error types are occurring, and the decomposition of the result overcomes the limitation (from whole to subdivision) caused by using the classification accuracy only. The confusion matrix is shown in table 1 below, positive representing a Positive class, i.e., a photovoltaic panel object, negative representing a non-photovoltaic panel object, TP representing the number of objects that predict a photovoltaic panel object as a photovoltaic panel object, FP representing the number of objects that predict a non-photovoltaic panel object as a photovoltaic panel object, FN representing the number of objects that predict a photovoltaic panel object as a non-photovoltaic panel object, and TN representing the number of objects that predict a non-photovoltaic panel object as a non-photovoltaic panel object; from this confusion matrix, the accuracy ACC and the omission factor can be calculated as shown in table 2 below.
TABLE 1
TABLE 2
The loss function in yolov5 is composed of two parts of loss of positive samples and negative samples, the positive samples are targets to be detected, the negative samples correspond to the background of the image, and if the negative samples are far more than the positive samples, the negative samples can submerge the loss of the positive samples, so that the network convergence efficiency and the detection precision are reduced. Through continuous testing, the similarity degree of the two frames is determined by adjusting down the category confidence threshold value confThreshold and reducing the IoU threshold value (IoU means that the intersection area of the real target frame and the predicted target frame is divided by the intersection area of the real target frame and the predicted target frame, and the greater the ratio, the higher the similarity is), statistics is carried out, and the accurate detection of the components can be realized for all test sets, for example, the detection omission ratio can be reduced to 0.012.
In some embodiments, as shown in fig. 8, edge coloring the photovoltaic module region and edge segmentation based on the edge coloring result includes:
and S61, a second preset color is formed on the edge of the photovoltaic module area in the photovoltaic panel area image according to the second coordinate information.
Specifically, the second preset color may be blue.
And S62, performing color detection on the colored photovoltaic panel area image to extract the outline of the photovoltaic module.
S63, carrying out contour recognition on the photovoltaic module, and carrying out edge segmentation according to the recognized contour.
In some embodiments, as shown in fig. 9, according to the edge segmentation result, the first coordinate information and the second coordinate information, a photovoltaic module segmentation result of the infrared image of the photovoltaic panel is obtained, including:
and S71, performing expansion and/or corrosion operation on the edge segmentation result, and updating the second coordinate information according to the operation result.
In this embodiment, the target detection frame of the photovoltaic module is rotated to frame other module areas, if the adjacent module is colored, the color extraction and the contour detection segmentation of the module are interfered, and at this time, the background color of the module is assigned with (0, 0), so that the interference of the segmentation of the adjacent module caused by the problem of the rotation angle can be avoided; because the convergence effect of the model cannot reach 100%, errors of edge segmentation can occur, the operations of expansion and corrosion are increased, meanwhile, the interference of small communication areas is eliminated, and finally, the accurate segmentation of the assembly can be realized, wherein the corrosion and expansion element structures can adopt convolution kernels of 3*3.
And S72, mapping the updated second coordinate information to the infrared image of the photovoltaic panel according to the first coordinate information to obtain a photovoltaic module segmentation result of the infrared image of the photovoltaic panel.
The result of the photovoltaic module segmentation is shown in fig. 10.
Specifically, regional coloring is carried out on the divided photovoltaic modules, the coloring color can be blue, then color detection (defining a blue threshold value) is carried out on each colored photovoltaic module part, and the outline of the coloring module is extracted; and then carrying out contour recognition on the component, and drawing out the contour of the component to carry out edge segmentation.
It should be noted that, in the above process, a problem may occur that the target detection frame of the component is rotated to frame another component area, and if the adjacent component is already colored, the color extraction and the contour detection segmentation of the component may be disturbed. For this purpose, the operation of assigning a value (0, 0) to the background of the component can be used, so that the interference of adjacent component segmentation caused by the problem of rotation angle is avoided.
Because the convergence effect of the model cannot reach 100%, errors of edge segmentation can occur, expansion operation and corrosion operation are increased, and meanwhile, interference of small communication areas is eliminated, so that accurate segmentation of the assembly can be realized. The corrosion operation and the expansion operation are described as follows:
the erosion operation erodes the boundary of the foreground object and removes small scale detail from the image, but at the same time reduces the size of the region of interest. In this operation, an odd-sized convolution kernel of arbitrary shape in the image is convolved, and if all pixels under the kernel are 1, then the pixel (1 or 0) in the original image is considered to be 1, otherwise it is eroded, even if it is zero. Thus, depending on the size of the kernel, all pixels near the boundary will be discarded, the thickness or size of the foreground object is reduced or the white area in the image is reduced. This operation is mainly used to remove small white noise. The structure size of the corrosion element can adopt a convolution kernel of 3*3.
The expansion operation is performed after the etching operation in order to remove noise. Since the etching operation removes white noise but also reduces the object, it is necessary to expand the object. Similar to the convolution operation, assuming that there are an image a and a structure element B, the structure element B moves over the image a, wherein the structure element B defines its center as an anchor point, and the maximum pixel value of a under B coverage is calculated to replace the pixel of the anchor point, wherein B may be any shape as a structure.
As described above, the dilation operation or erosion operation is to convolve an image (or a portion of an area of an image, referred to as image a) with a structuring element, referred to as a convolution kernel B. The core may be of any shape and size, having a separately defined reference point, known as an anchor point (anchor). In most cases, the kernel is a small solid square or disk with a reference point in the middle, which can be considered a template or mask. Dilation is the operation of taking a local maximum, the kernel B convolving with the pattern, i.e. calculating the maximum of the pixel points of the area covered by the kernel B, and assigning this maximum to the pixel specified by the reference point. This causes the highlight region in the image to grow gradually) the 3*3 size of the dilated element structure can be used.
In some embodiments, as shown in fig. 11, the method further comprises:
s91, generating a background image, wherein the size of the background image is the same as that of the infrared image of the photovoltaic panel, and the color is a first preset color.
The first preset color may be black.
And S92, generating a mask image of the photovoltaic module in the infrared image of the photovoltaic panel according to the first coordinate information, the second coordinate information and the background image.
Specifically, an image with black background and the same size as the original image can be duplicated, and the information of the extracted contour coordinate points is mapped into the image, so as to generate a mask image, wherein the mask image is shown in fig. 12.
In summary, according to the method for segmenting the infrared image of the photovoltaic panel, disclosed by the embodiment of the invention, the photovoltaic panel image is segmented, and then the photovoltaic module is segmented from the photovoltaic panel image by using the example segmentation model added with the small target detection layer, so that the time cost is reduced, and the segmentation precision is improved.
Based on the method for segmenting the infrared image of the photovoltaic panel in the embodiment, the invention further provides a computer readable storage medium.
In this embodiment, a computer program is stored on a computer readable storage medium, and when the computer program is executed by a processor, the above-mentioned method for dividing an infrared image of a photovoltaic panel is implemented.
Fig. 13 is a block diagram of an electronic device according to an embodiment of the invention.
As shown in fig. 13, the electronic device 100 includes: a processor 101 and a memory 103. Wherein the processor 101 is coupled to the memory 103, such as via bus 102. Optionally, the electronic device 100 may also include a transceiver 104. It should be noted that, in practical applications, the transceiver 104 is not limited to one, and the structure of the electronic device 100 is not limited to the embodiment of the present invention.
The processor 101 may be a CPU (Central Processing Unit ), general purpose processor, DSP (Digital Signal Processor, data signal processor), ASIC (Application Specific Integrated Circuit ), FPGA (Field Programmable Gate Array, field programmable gate array) or other programmable logic device, transistor logic device, hardware components, or any combination thereof. Which may implement or perform the various exemplary logical blocks, modules, and circuits described in connection with the present disclosure. The processor 101 may also be a combination that implements computing functionality, e.g., comprising one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
Bus 102 may include a path to transfer information between the aforementioned components. Bus 102 may be a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus or an EISA (Extended Industry Standard Architecture ) bus, or the like. The bus 102 may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, only one thick line is shown in fig. 13, but not only one bus or one type of bus.
The memory 103 is used for storing a computer program corresponding to the photovoltaic panel infrared image segmentation method of the above-described embodiment of the present invention, which is controlled to be executed by the processor 101. The processor 101 is configured to execute a computer program stored in the memory 103 to implement what is shown in the foregoing method embodiments.
Among other things, the electronic device 100 includes, but is not limited to: mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and the like, and stationary terminals such as digital TVs, desktop computers, and the like. The electronic device 100 shown in fig. 13 is only an example and should not be construed as limiting the functionality and scope of use of the embodiments of the invention.
It should be noted that the logic and/or steps represented in the flowcharts or otherwise described herein, for example, may be considered as a ordered listing of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the device or element being referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," "secured," and the like are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; either directly or indirectly, through intermediaries, or both, may be in communication with each other or in interaction with each other, unless expressly defined otherwise. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In the present invention, unless expressly stated or limited otherwise, a first feature "up" or "down" a second feature may be the first and second features in direct contact, or the first and second features in indirect contact via an intervening medium. Moreover, a first feature being "above," "over" and "on" a second feature may be a first feature being directly above or obliquely above the second feature, or simply indicating that the first feature is level higher than the second feature. The first feature being "under", "below" and "beneath" the second feature may be the first feature being directly under or obliquely below the second feature, or simply indicating that the first feature is less level than the second feature.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (8)
1. A method for segmenting an infrared image of a photovoltaic panel, the method comprising:
acquiring a photovoltaic panel infrared image shot by an unmanned aerial vehicle;
performing convolution operation on the infrared image of the photovoltaic panel to position the photovoltaic panel in the infrared image of the photovoltaic panel;
fitting the positioning result based on the color characteristics and the shape characteristics of the photovoltaic panel to optimize the positioning result;
obtaining a photovoltaic panel area image and first coordinate information of a photovoltaic panel area in the photovoltaic panel infrared image according to an optimization result;
inputting the photovoltaic panel region image into a pre-trained example segmentation model based on yolov5 to obtain a photovoltaic module region and second coordinate information of the photovoltaic module region in the photovoltaic panel region image, wherein the example segmentation model based on yolov5 is provided with at least four groups of anchor parameters, and at least one small target detection layer is added;
performing edge coloring on the photovoltaic module area, and performing edge segmentation based on an edge coloring result;
performing expansion and/or corrosion operation on the edge segmentation result, and updating the second coordinate information according to the operation result;
and mapping the updated second coordinate information to the infrared image of the photovoltaic panel according to the first coordinate information to obtain a photovoltaic module segmentation result of the infrared image of the photovoltaic panel.
2. The method of photovoltaic panel infrared image segmentation according to claim 1, further comprising:
generating a background image, wherein the size of the background image is the same as that of the infrared image of the photovoltaic panel, and the color is a first preset color;
and generating a mask image of the photovoltaic module in the infrared image of the photovoltaic panel according to the first coordinate information, the second coordinate information and the background image.
3. The method of claim 1, wherein the training process based on the yolov5 example segmentation model comprises:
acquiring a training sample set, wherein the training sample set comprises a positive sample and a negative sample, the positive sample is an infrared image containing a photovoltaic panel, and the negative sample is an infrared image not containing the photovoltaic panel;
dividing the photovoltaic panel in the positive sample to obtain a photovoltaic panel image, and marking the photovoltaic module in the photovoltaic panel image;
dividing the negative sample and the marked photovoltaic panel image into a training set and a testing set;
and constructing the example segmentation model based on the yolov5, training the example segmentation model based on the yolov5 by utilizing the training set, and testing the trained example segmentation model based on the yolov5 by utilizing the testing set to obtain the trained example segmentation model based on the yolov 5.
4. A method of photovoltaic panel infrared image segmentation according to claim 3, characterized in that the trained yolov 5-based instance segmentation model is tested by lowering the category confidence threshold and IoU threshold multiple times during the training of the yolov 5-based instance segmentation model with the test set.
5. The method for segmenting an infrared image of a photovoltaic panel according to claim 1, wherein the step of edge-coloring the photovoltaic module region and edge-segmenting based on the result of edge-coloring comprises:
according to the second coordinate information, a second preset color is formed on the edge of the photovoltaic module area in the photovoltaic panel area image;
performing color detection on the colored photovoltaic panel area image to extract the outline of the photovoltaic module;
and carrying out contour recognition on the photovoltaic module, and carrying out edge segmentation according to the recognized contour.
6. The method of any one of claims 1-5, wherein the yolov 5-based example segmentation model is provided with four sets of anchor parameters, respectively [5,6;8,14;15,11] #4, [10,13;16,30;33,23] # P3/8, [30,61;62,45;59,119] # P4/16, [116,90;156,198;373,326] # P5/32.
7. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of segmentation of infrared images of a photovoltaic panel according to any one of claims 1-6.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory, which when executed by the processor, implements the method of infrared image segmentation of a photovoltaic panel as claimed in any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310469313.0A CN116188502B (en) | 2023-04-27 | 2023-04-27 | Method for dividing infrared image of photovoltaic panel, storage medium and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310469313.0A CN116188502B (en) | 2023-04-27 | 2023-04-27 | Method for dividing infrared image of photovoltaic panel, storage medium and electronic device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116188502A CN116188502A (en) | 2023-05-30 |
| CN116188502B true CN116188502B (en) | 2023-07-21 |
Family
ID=86452625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310469313.0A Active CN116188502B (en) | 2023-04-27 | 2023-04-27 | Method for dividing infrared image of photovoltaic panel, storage medium and electronic device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116188502B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021149857A1 (en) * | 2020-01-22 | 2021-07-29 | 고려대학교 세종산학협력단 | Accurate animal detection method and apparatus using yolo-based light-weight bounding box detection and image processing |
| CN113989241A (en) * | 2021-10-29 | 2022-01-28 | 南京埃斯顿机器人工程有限公司 | Photovoltaic module EL defect detection method based on image processing and deep learning fusion |
| WO2022151755A1 (en) * | 2021-01-15 | 2022-07-21 | 上海商汤智能科技有限公司 | Target detection method and apparatus, and electronic device, storage medium, computer program product and computer program |
| CN114841932A (en) * | 2022-04-18 | 2022-08-02 | 上海电气集团股份有限公司 | Foreign matter detection method, system, equipment and medium for photovoltaic panel of photovoltaic power station |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113538453A (en) * | 2021-05-27 | 2021-10-22 | 旻投电力发展有限公司 | Infrared image photovoltaic module visual identification method based on semantic segmentation |
| CN113538391B (en) * | 2021-07-25 | 2022-08-02 | 吉林大学 | Photovoltaic defect detection method based on Yolov4 and thermal infrared image |
| CN113553977B (en) * | 2021-07-30 | 2023-02-10 | 国电汉川发电有限公司 | Improved YOLO V5-based safety helmet detection method and system |
| CN115294031A (en) * | 2022-07-18 | 2022-11-04 | 青海黄河上游水电开发有限责任公司光伏产业技术分公司 | Photovoltaic module fault image identification method based on infrared thermal imaging analysis |
| CN115797336A (en) * | 2023-02-01 | 2023-03-14 | 尚特杰电力科技有限公司 | Fault detection method and device of photovoltaic module, electronic equipment and storage medium |
-
2023
- 2023-04-27 CN CN202310469313.0A patent/CN116188502B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021149857A1 (en) * | 2020-01-22 | 2021-07-29 | 고려대학교 세종산학협력단 | Accurate animal detection method and apparatus using yolo-based light-weight bounding box detection and image processing |
| WO2022151755A1 (en) * | 2021-01-15 | 2022-07-21 | 上海商汤智能科技有限公司 | Target detection method and apparatus, and electronic device, storage medium, computer program product and computer program |
| CN113989241A (en) * | 2021-10-29 | 2022-01-28 | 南京埃斯顿机器人工程有限公司 | Photovoltaic module EL defect detection method based on image processing and deep learning fusion |
| CN114841932A (en) * | 2022-04-18 | 2022-08-02 | 上海电气集团股份有限公司 | Foreign matter detection method, system, equipment and medium for photovoltaic panel of photovoltaic power station |
Non-Patent Citations (2)
| Title |
|---|
| 基于图像处理的光伏组件串识别技术研究;徐伟;张侃健;;信息技术与信息化(01);193-196 * |
| 远红外车载图像实时行人检测与自适应实例分割;于博;马书浩;李红艳;李春庚;安居白;;激光与光电子学进展(02);1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116188502A (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108009543B (en) | License plate recognition method and device | |
| CN108288088B (en) | Scene text detection method based on end-to-end full convolution neural network | |
| CN107944450B (en) | License plate recognition method and device | |
| EP3806064B1 (en) | Method and apparatus for detecting parking space usage condition, electronic device, and storage medium | |
| CN109978839B (en) | Method for detecting wafer low-texture defects | |
| CN109977997B (en) | Image target detection and segmentation method based on convolutional neural network rapid robustness | |
| CN107239794B (en) | Point cloud data segmentation method and terminal | |
| CN111209920B (en) | Airplane detection method under complex dynamic background | |
| CN112560698B (en) | Image processing method, device, equipment and medium | |
| CN111914720B (en) | Method and device for identifying insulator burst of power transmission line | |
| CN110176042A (en) | Training method, device and the storage medium of camera self moving parameter estimation model | |
| CN111597941B (en) | Target detection method for dam defect image | |
| CN115439840A (en) | Aviation piece slot area identification method, device, equipment and medium | |
| CN114549393B (en) | Image labeling method, device, equipment and computer storage medium | |
| CN114820679A (en) | Image annotation method and device, electronic equipment and storage medium | |
| CN113705564B (en) | Pointer type instrument identification reading method | |
| CN116188502B (en) | Method for dividing infrared image of photovoltaic panel, storage medium and electronic device | |
| CN113284185A (en) | Rotating target detection method for remote sensing target detection | |
| CN110817674B (en) | Method, device and equipment for detecting step defect of escalator and storage medium | |
| CN112634259A (en) | Automatic modeling and positioning method for keyboard keycaps | |
| CN112528994A (en) | Free-angle license plate detection method, license plate identification method and identification system | |
| CN113840135B (en) | Color cast detection method, device, equipment and storage medium | |
| CN116993654A (en) | Camera module defect detection method, device, equipment, storage medium and product | |
| CN112241697B (en) | Corner color determination method and device, terminal device and readable storage medium | |
| CN111709377B (en) | Feature extraction method, target re-identification method and device and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |