CN114898367A - Intelligent identification method, device, equipment and medium for electric meter dial numbers - Google Patents
Intelligent identification method, device, equipment and medium for electric meter dial numbersDownload PDFInfo
- Publication number
- CN114898367A CN114898367ACN202210549315.6ACN202210549315ACN114898367ACN 114898367 ACN114898367 ACN 114898367ACN 202210549315 ACN202210549315 ACN 202210549315ACN 114898367 ACN114898367 ACN 114898367A
- Authority
- CN
- China
- Prior art keywords
- image
- dial
- electric meter
- meter
- predicted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/1444—Selective acquisition, locating or processing of specific regions, e.g. highlighted text, fiducial marks or predetermined fields
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Character Discrimination (AREA)
Abstract
Translated fromChinese
Description
Translated fromChinese技术领域technical field
本发明涉及人工智能领域，尤其涉及一种电表表盘数字的智能识别方法、装置、电子设备以及计算机可读存储介质。The present invention relates to the field of artificial intelligence, and in particular, to an intelligent identification method, device, electronic device and computer-readable storage medium for numbers on an electric meter dial.
背景技术Background technique
电表是用来过测量电能的仪表，又称电度表，火表，电能表，千瓦小时表，指测量各种电学量的仪表，其应用于不同的生活领域中，如所述电表可以用于统计居民日常生活的用电状态，因此，如何高效读取的电表中的数字显得愈发重要。Electricity meter is an instrument used to measure electric energy, also known as watt-hour meter, fire meter, electric energy meter, kilowatt-hour meter, which refers to an instrument that measures various electrical quantities. It is used in different life fields. In order to count the electricity consumption status of residents in daily life, how to efficiently read the figures in the electricity meter becomes more and more important.
目前，电表的数值多数仍旧依赖人工的方式，即通过抄表员上门抄表读取数值，这样不仅有人工的成本，在一定程度上影响了电表数值读取的时效性，因此，亟待一种电表数字的自动识别方法，以提高电表数字的读取效率。At present, most of the values of the electric meter still rely on manual methods, that is, the reading of the value through the door-to-door meter reading by the meter reader, which not only has labor costs, but also affects the timeliness of reading the value of the electric meter to a certain extent. The automatic identification method of the electric meter number to improve the reading efficiency of the electric meter number.
发明内容SUMMARY OF THE INVENTION
为了解决上述技术问题，本发明提供了一种电表表盘数字的智能识别方法、装置、电子设备以及计算机可读存储介质，可以实现电表数字的自动识别方法，提高电表数字的读取效率。In order to solve the above technical problems, the present invention provides an intelligent identification method, device, electronic device and computer-readable storage medium for meter dial numbers, which can realize the automatic identification method of meter numbers and improve the reading efficiency of meter numbers.
第一方面，本发明提供了一种电表表盘数字的智能识别方法，包括：In a first aspect, the present invention provides a method for intelligently identifying numbers on an electric meter dial, including:
采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字；Collect the dial image of the historical electricity meter, perform a preprocessing operation on the dial image, obtain the target dial image, and mark the real electricity meter area and the real electricity meter number of the target dial image;
将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字；The target dial image is input into a pre-built image digital recognition model to detect the predicted meter area of the target dial image through the area detection network in the image digital recognition model, and use the image digital recognition model to detect the predicted meter area. The text recognition network identifies the predicted meter number in the predicted meter area;
根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失；Calculate the model loss of the image number recognition model according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the real electric meter number;
判断所述模型损失是否满足预设条件；Determine whether the model loss satisfies a preset condition;
若所述模型损失不满足所述预设条件，调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤；If the model loss does not meet the preset condition, adjust the parameters of the image number recognition model, and return to executing the step of inputting the target dial image into the pre-built image number recognition model;
若所述模型损失满足所述预设条件，得到训练完成的图像数字识别模型；If the model loss satisfies the preset condition, a trained image digital recognition model is obtained;
利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。Using the trained image number recognition model to perform digital recognition on the dial image of the current electric meter to obtain the dial number of the current electric meter.
可以看出，本发明实施例首先对历史电表的表盘图像进行预处理操作，得到目标表盘图像，可以减少后续模型的训练无用数据，提高模型训练的速度和质量，并标注目标表盘图像的真实电表区域及真实电表数字，可以作为后续模型训练结果的数据参照，监督模型的学习，提高模型的识别能力；其次，本发明实施例通过将目标表盘图像输入至预构建的图像数字识别模型中，以通过目标表盘图像对图像数字识别模型中的区域检测网络和文字识别网络进行训练，得到训练完成的图像数字识别模型，以减少后续人为参与电表数字的读取工作，提高后续电表的读取效率；进一步地，本发明实施例利用训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到当前电表的表盘数字，实现当前电表的数字智能读取。因此，本发明实施例提出的一种电表表盘数字的智能识别方法可以实现电表数字的自动识别方法，提高电表数字的读取效率。It can be seen that the embodiment of the present invention first performs a preprocessing operation on the dial image of the historical electricity meter to obtain the target dial image, which can reduce the useless data of subsequent model training, improve the speed and quality of model training, and mark the real electricity meter of the target dial image. The area and the real meter number can be used as a data reference for the subsequent model training results to supervise the learning of the model and improve the recognition ability of the model; The region detection network and text recognition network in the image digital recognition model are trained through the target dial image, and the trained image digital recognition model is obtained, so as to reduce the subsequent manual participation in the reading of the meter numbers and improve the reading efficiency of the subsequent meters; Further, the embodiment of the present invention uses the trained image digital recognition model to perform digital recognition on the dial image of the current electric meter, obtains the dial number of the current electric meter, and realizes the digital intelligent reading of the current electric meter. Therefore, the intelligent identification method for the numbers on the dial of the electric meter proposed by the embodiment of the present invention can realize the automatic identification method of the numbers of the electric meter, and improve the reading efficiency of the numbers of the electric meter.
在第一方面的一种可能实现方式中，所述对所述表盘图像进行预处理操作，得到目标表盘图像包括：In a possible implementation manner of the first aspect, performing a preprocessing operation on the dial image to obtain the target dial image includes:
检测所述表盘图像中是否存在损伤图像，若所述表盘图像中存在损伤图像，将所述损伤图像剔除后检测所述表盘图像中是否存在重复图像；Detecting whether there is a damaged image in the dial image, and if there is a damaged image in the dial image, remove the damaged image and detect whether there is a duplicate image in the dial image;
若所述表盘图像中存在重复图像，将所述重复图像剔除后生成目标表盘图像。If there is a duplicate image in the dial image, a target dial image is generated after removing the duplicate image.
在第一方面的一种可能实现方式中，所述检测所述表盘图像中是否存在损伤图像，包括：In a possible implementation manner of the first aspect, the detecting whether there is a damaged image in the dial image includes:
计算所述表盘图像中每个图像的损伤值；calculating a damage value for each of the dial images;
若所述损伤值大于预设阈值，将所述图像作为损伤图像；If the damage value is greater than a preset threshold, use the image as a damage image;
若所述损伤值不大于所述预设阈值，将所述图像作为未损伤图像。If the damage value is not greater than the preset threshold, the image is regarded as an undamaged image.
在第一方面的一种可能实现方式中，所述通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，包括：In a possible implementation manner of the first aspect, the detecting the predicted electric meter area of the target dial image through the area detection network in the image digital recognition model includes:
利用所述区域检测网络中的卷积层对所述目标表盘图像进行图像特征提取，得到特征表盘图像；Use the convolution layer in the area detection network to perform image feature extraction on the target dial image to obtain a characteristic dial image;
利用所述区域检测网络中的批标准化层对所述特征表盘图像进行标准化操作，得到标准表盘图像；Use the batch normalization layer in the area detection network to standardize the feature dial image to obtain a standard dial image;
利用所述区域检测网络中的激活函数输出所述标准表盘图像的检测结果，以生成所述目标表盘图像的预测电表区域。The detection result of the standard dial image is output by using the activation function in the area detection network, so as to generate the predicted electric meter area of the target dial image.
在第一方面的一种可能实现方式中，所述利用所述区域检测网络中的批标准化层对所述特征表盘图像进行标准化操作，得到标准表盘图像，包括：In a possible implementation manner of the first aspect, the standardization operation is performed on the characteristic dial image by using the batch normalization layer in the area detection network to obtain a standard dial image, including:
利用下述公式对所述特征表盘图像进行标准化操作：The characteristic dial image is standardized by the following formula:
其中，x′i为标准表盘图像，xi为特征表盘图像，μ为特征表盘图像的均值，σ2为特征表盘图像的方差，ε为无穷小的随机数。Among them, x′i is the standard dial image,xi is the characteristic dial image, μ is the mean value of the characteristic dial image, σ2 is the variance of the characteristic dial image, and ε is an infinitesimal random number.
在第一方面的一种可能实现方式中，所述利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字，包括：In a possible implementation manner of the first aspect, the use of a character recognition network in the image number recognition model to identify the predicted electric meter number in the predicted electric meter area includes:
利用所述文字识别网络中的输入门计算所述预测电表区域的状态值；Using the input gate in the character recognition network to calculate the state value of the predicted meter area;
利用所述文字识别网络中的遗忘门计算所述预测电表区域的激活值；Calculate the activation value of the predicted meter area by using the forget gate in the character recognition network;
根据所述状态值和激活值计算所述预测电表区域的状态更新值；calculating a state update value for the predicted meter area according to the state value and the activation value;
利用所述文字识别网络中的输出门计算所述状态更新值的数字序列，得到预测电表数字。The digital sequence of the state update value is calculated by using the output gate in the character recognition network to obtain the predicted meter number.
8、在第一方面的一种可能实现方式中，所述根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失，包括：8. In a possible implementation manner of the first aspect, according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the real electric meter number, calculate the image number recognition model. Model losses, including:
根据所述真实电表区域和所述预测电表区域，计算所述图像数字识别模型的第一损失；calculating the first loss of the image digital recognition model according to the real meter area and the predicted meter area;
根据所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的第二损失；calculating a second loss of the image number recognition model according to the predicted meter number and the real meter number;
根据所述第一损失和所述第二损失，计算所述图像数字识别模型的模型损失。Based on the first loss and the second loss, a model loss of the image digit recognition model is calculated.
第二方面，本发明提供了一种电表表盘数字的智能识别装置，所述装置包括：In a second aspect, the present invention provides an intelligent identification device for numbers on an electric meter dial, the device comprising:
图像预处理模块，用于采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字；The image preprocessing module is used to collect the dial image of the historical electric meter, perform a preprocessing operation on the dial image, obtain the target dial image, and mark the real electric meter area and the real electric meter number of the target dial image;
模型训练模块，用于将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字；The model training module is used to input the target dial image into the pre-built image digital recognition model, so as to detect the predicted meter area of the target dial image through the area detection network in the image digital recognition model, and use the The text recognition network in the image number recognition model recognizes the predicted meter number in the predicted meter area;
所述模型训练模块，还用于根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失；The model training module is further configured to calculate the model loss of the image number recognition model according to the real meter area and the predicted meter area, and the predicted meter number and the real meter number;
所述模型训练模块，还用于判断所述模型损失是否满足预设条件；The model training module is further configured to judge whether the model loss satisfies a preset condition;
所述模型训练模块，还用于在所述模型损失不满足所述预设条件时，调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤；The model training module is further configured to adjust the parameters of the image digital recognition model when the model loss does not meet the preset conditions, and return to executing the inputting the target dial image into the pre-built image steps in the digital recognition model;
所述模型训练模块，还用于在所述模型损失满足所述预设条件时，得到训练完成的图像数字识别模型；The model training module is further configured to obtain a trained image digital recognition model when the model loss satisfies the preset condition;
电表数字识别模块，用于利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。The electric meter digital recognition module is used for digitally recognizing the dial image of the current electric meter by using the image digital recognition model completed by the training to obtain the dial number of the current electric meter.
第三方面，本发明提供一种电子设备，包括：In a third aspect, the present invention provides an electronic device, comprising:
至少一个处理器；以及与所述至少一个处理器通信连接的存储器；at least one processor; and a memory communicatively coupled to the at least one processor;
其中，所述存储器存储有可被所述至少一个处理器执行的计算机程序，以使所述至少一个处理器能够执行如上述第一方面中任意一项所述的电表表盘数字的智能识别方法。Wherein, the memory stores a computer program executable by the at least one processor, so that the at least one processor can execute the method for intelligently recognizing numbers on an electric meter dial according to any one of the first aspect above.
第四方面，本发明提供一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时实现如上述第一方面中任意一项所述的电表表盘数字的智能识别方法。In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the method for intelligently recognizing numbers on a meter dial according to any one of the first aspects above is implemented.
可以理解的是，上述第二方面至第四方面的有益效果可以参见上述第一方面中的相关描述，在此不再赘述。It can be understood that, for the beneficial effects of the foregoing second aspect to the fourth aspect, reference may be made to the relevant descriptions in the foregoing first aspect, and details are not described herein again.
附图说明Description of drawings
此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本发明的实施例，并与说明书一起用于解释本发明的原理。The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description serve to explain the principles of the invention.
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，对于本领域普通技术人员而言，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. In other words, on the premise of no creative labor, other drawings can also be obtained from these drawings.
图1为本发明一实施例提供的一种电表表盘数字的智能识别方法的流程示意图；1 is a schematic flowchart of a method for intelligently identifying numbers on an electric meter dial provided by an embodiment of the present invention;
图2为本发明一实施例中图1提供的一种电表表盘数字的智能识别方法的其中一个步骤的流程示意图；FIG. 2 is a schematic flowchart of one step of a method for intelligently identifying numbers on an electric meter dial provided in FIG. 1 according to an embodiment of the present invention;
图3为本发明一实施例中图1提供的一种电表表盘数字的智能识别方法的另外一个步骤的流程示意图；FIG. 3 is a schematic flowchart of another step of the method for intelligently identifying numbers on an electric meter dial provided in FIG. 1 according to an embodiment of the present invention;
图4为本发明一实施例提供的一种电表表盘数字的智能识别装置的模块示意图；4 is a schematic block diagram of an intelligent identification device for numbers on an electric meter dial provided by an embodiment of the present invention;
图5为本发明一实施例提供的实现电表表盘数字的智能识别方法的电子设备的内部结构示意图。FIG. 5 is a schematic diagram of an internal structure of an electronic device for implementing an intelligent identification method for numbers on an electric meter dial according to an embodiment of the present invention.
具体实施方式Detailed ways
为使本发明实施例的目的、技术方案和优点更加清楚，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本发明的一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本发明保护的范围。In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present invention.
参阅图1所示，是本发明一实施例提供的电表表盘数字的智能识别方法的流程示意图。其中，图1中描述的电表表盘数字的智能识别方法包括：Referring to FIG. 1 , it is a schematic flowchart of a method for intelligently recognizing numbers on an electric meter dial provided by an embodiment of the present invention. Among them, the intelligent identification method of the number of the meter dial described in FIG. 1 includes:
S1、采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字。S1. Collect a dial image of a historical electric meter, perform a preprocessing operation on the dial image, obtain a target dial image, and mark the real electric meter area and real electric meter number of the target dial image.
本发明实施例中，所述历史电表是指经过测量电能的仪表，其基于不同的电表属性可以划分多种类型的电表，如按照电表用途可以划分为工业与民用表、电子标准表、最大需量表以及复费率表等，按照电表的接入电源性质可以划分为交流表、直流表等，按照电表的用电设备可以划分为单相、三相三线、三相四线电能表等，所述表盘图像是指用于读取所述历史电表的电能数值区域图像，在本发明中，所述表盘图像可以通过在所述历史电表的表盘正上方安装摄像头采集得到，以保障采集的表盘图像中的表盘处于正方向，提高后续表盘数字识别的准确率。In the embodiment of the present invention, the historical electricity meter refers to a meter that has measured electric energy, which can be divided into various types of electricity meters based on different electricity meter attributes. Meters and multi-rate meters, etc., can be divided into AC meters, DC meters, etc. according to the nature of the access power of the meters, and can be divided into single-phase, three-phase three-wire, three-phase four-wire energy meters, etc. according to the electrical equipment of the meter. The dial image refers to the image of the electric energy value area used to read the historical electricity meter. In the present invention, the dial image can be collected by installing a camera directly above the dial of the historical electricity meter, so as to ensure the collected dial. The dial in the image is in the positive direction, which improves the accuracy of subsequent dial number recognition.
应该了解，由于业务场景的错综复杂，导致采集的表盘图像会存在损伤或者重复的现象，因此，本发明实施例通过对所述表盘图像进行预处理操作，以筛选出所述表盘图像中的一些无用图像，减少后续模型的训练无用数据，提高模型训练的速度和质量。It should be understood that due to the intricate business scenarios, the collected dial images may be damaged or repeated. Therefore, in the embodiment of the present invention, the dial images are preprocessed to filter out some useless dial images. Images, reduce useless data for subsequent model training, and improve the speed and quality of model training.
作为本发明的一个实施例，所述对所述表盘图像进行预处理操作，得到目标表盘图像包括：检测所述表盘图像中是否存在损伤图像，若所述表盘图像中存在损伤图像，将所述损伤图像剔除后检测所述表盘图像中是否存在重复图像，若所述表盘图像中存在重复图像，将所述重复图像剔除后生成目标表盘图像。需要说明的是，若所述表盘图像中既不存在损伤图像又不存在重复图像，则将所述表盘图像作为所述目标表盘图像。As an embodiment of the present invention, performing a preprocessing operation on the dial image to obtain the target dial image includes: detecting whether there is a damaged image in the dial image, and if there is a damaged image in the dial image, converting the dial image to the target dial image. After the damaged image is removed, it is detected whether there is a duplicate image in the dial image, and if there is a duplicate image in the dial image, a target dial image is generated after removing the duplicate image. It should be noted that, if there is neither a damaged image nor a repeated image in the dial image, the dial image is used as the target dial image.
进一步地，本发明一可选实施例中，所述检测所述表盘图像中是否存在损伤图像，包括：计算所述表盘图像中每个图像的损伤值，若所述损伤值大于预设阈值，将所述图像作为损伤图像，若所述损伤值不大于所述预设阈值，将所述图像作为未损伤图像。其中，所述损失值可以设置为0.1，也可以根据实际业务场景设置。Further, in an optional embodiment of the present invention, the detecting whether there is a damaged image in the dial image includes: calculating a damage value of each image in the dial image, if the damage value is greater than a preset threshold, The image is regarded as a damaged image, and if the damage value is not greater than the preset threshold, the image is regarded as an undamaged image. The loss value may be set to 0.1, or may be set according to actual business scenarios.
进一步地，本发明又一可选实施例中，利用下述公式计算所述表盘图像中每个图像的损伤值：Further, in another optional embodiment of the present invention, the following formula is used to calculate the damage value of each image in the dial image:
γ＝αln(b+1)γ=αln(b+1)
其中，Lb(x)表示损伤值，x表示表盘图像中图像的像素值，α和b分别表示表盘图像中图像的权重和偏置，C表示表盘图像中图像的归一化参数。Among them, Lb (x) represents the damage value, x represents the pixel value of the image in the dial image, α and b represent the weight and bias of the image in the dial image, respectively, and C represents the normalization parameter of the image in the dial image.
进一步地，本发明一可选实施例中，所述表盘图像的重复图像检测可以通过图像检测算法实现，如OpenCV算法。Further, in an optional embodiment of the present invention, the repeated image detection of the dial image may be implemented by an image detection algorithm, such as an OpenCV algorithm.
进一步地，本发明实施例通过标注所述目标表盘图像的真实电表区域及真实电表数字，以作为后续模型训练结果的数据参照，监督模型的学习，提高模型的识别能力，其中，所述真实电表区域是指在所述表盘图像中数字所在区域，所述真实电表数字是指在所述表盘图像中的电表数值。Further, the embodiment of the present invention supervises the learning of the model and improves the recognition ability of the model by marking the real electric meter area and the real electric meter number of the target dial image as a data reference for the subsequent model training results, wherein the real electric meter The area refers to the area where the numbers are located in the dial image, and the real meter number refers to the meter value in the dial image.
S2、将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字。S2. Input the target dial image into a pre-built image digital recognition model, so as to detect the predicted meter area of the target dial image through the area detection network in the image digital recognition model, and use the image digital recognition A text recognition network in the model identifies predicted meter numbers for the predicted meter area.
本发明实施例中，所述预构建的图像数字识别模型包括区域检测网络和文字识别网络，所述区域检测网络可以通过YOLO算法构建，其用于检测所述目标表盘图像中的电表区域，所述文字识别网络包括长短期记忆网络(Long Short-Term Memory，LSTM)，其用于对区域检测网络识别的电表区域进行电表数字读取。In the embodiment of the present invention, the pre-built image digital recognition model includes an area detection network and a character recognition network, and the area detection network can be constructed by the YOLO algorithm, which is used to detect the electric meter area in the target dial image, so The text recognition network includes a long short-term memory network (Long Short-Term Memory, LSTM), which is used to read the meter numbers for the meter area identified by the area detection network.
进一步地，本发明实施例在将所述目标表盘图像输入至预构建的图像数字识别模型中之前，还包括：统一所述目标表盘图像的图像格式，以将所述目标表盘图像转换成所述图像数字识别模型可以识别的格式图像，提高模型训练速度，可选的，所述目标表盘图像的图像格式统一可以通过CV2指令实现。Further, before inputting the target dial image into the pre-built image digital recognition model, the embodiment of the present invention further includes: unifying the image format of the target dial image, so as to convert the target dial image into the The format image that can be recognized by the image digital recognition model can improve the training speed of the model. Optionally, the image format of the target dial image can be unified through the CV2 instruction.
进一步，作为本发明的一个实施例，参阅图2所示，所述通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，包括：Further, as an embodiment of the present invention, referring to FIG. 2 , the detection of the predicted meter area of the target dial image through the area detection network in the image digital recognition model includes:
S201、利用所述区域检测网络中的卷积层对所述目标表盘图像进行图像特征提取，得到特征表盘图像；S201, using the convolution layer in the area detection network to perform image feature extraction on the target dial image to obtain a characteristic dial image;
S202、利用所述区域检测网络中的批标准化层对所述特征表盘图像进行标准化操作，得到标准表盘图像；S202, using the batch normalization layer in the area detection network to standardize the feature dial image to obtain a standard dial image;
S203、利用所述区域检测网络中的激活函数输出所述标准表盘图像的检测结果，以生成所述目标表盘图像的预测电表区域。S203. Use the activation function in the area detection network to output the detection result of the standard dial image, so as to generate the predicted meter area of the target dial image.
本发明一可选实施中，所述图像特征提取通过所述卷积层中的卷积核实现，所述批标准化层对提取的图像特征进行标准化，可以加速模型的收敛。In an optional implementation of the present invention, the image feature extraction is implemented by a convolution kernel in the convolution layer, and the batch normalization layer normalizes the extracted image features, which can speed up the convergence of the model.
本发明一可选实施例中，利用下述公式对所述特征表盘图像进行标准化操作：In an optional embodiment of the present invention, the following formula is used to standardize the characteristic dial image:
其中，x′i为标准表盘图像，xi为特征表盘图像，μ为特征表盘图像的均值，σ2为特征表盘图像的方差，ε为无穷小的随机数。Among them, x′i is the standard dial image,xi is the characteristic dial image, μ is the mean value of the characteristic dial image, σ2 is the variance of the characteristic dial image, and ε is an infinitesimal random number.
本发明一可选实施例中，所述激活函数包括：In an optional embodiment of the present invention, the activation function includes:
其中，s′表示激活后的电表，s表示标准表盘图像。Among them, s' represents the activated electric meter, and s represents the standard dial image.
本发明一可选实施例中，所述检测结果包括：x、y、高、宽以及类别等，其中，x、y表示目标特征图像的中心点，类别表示目标特征图像是否为电表区域，即类别0表示不是电表区域，类别1表示是电表区域，于是，本发明选取类别为1的电表作为预测电表区域。In an optional embodiment of the present invention, the detection result includes: x, y, height, width, category, etc., where x and y represent the center point of the target feature image, and the category represents whether the target feature image is the meter area, that is, Category 0 indicates that it is not an electric meter area, and category 1 indicates that it is an electric meter area. Therefore, the present invention selects an electric meter with category 1 as the predicted electric meter area.
进一步地，作为本发明的一个实施例，所述利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字，包括：利用所述文字识别网络中的输入门计算所述预测电表区域的状态值；利用所述文字识别网络中的遗忘门计算所述预测电表区域的激活值；根据所述状态值和激活值计算所述预测电表区域的状态更新值；利用所述文字识别网络中的输出门计算所述状态更新值的数字序列，得到预测电表数字。Further, as an embodiment of the present invention, the use of the character recognition network in the image number recognition model to identify the predicted electric meter number in the predicted electric meter area includes: using an input gate in the character recognition network to calculate the predicted electric meter number. the state value of the predicted electric meter area; use the forget gate in the character recognition network to calculate the activation value of the predicted electric meter area; calculate the state update value of the predicted electric meter area according to the state value and the activation value; use the The output gate in the text recognition network calculates the sequence of numbers of the state update value to obtain the predicted meter number.
S3、根据所述真实电表区域和所述预测电表区域，及所述预测电表数字及所述真实电表数字，计算所述图像数字识别模型的模型损失。S3. Calculate the model loss of the image number recognition model according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the real electric meter number.
本发明实施例中，参阅图3所示，所述根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失，包括：In the embodiment of the present invention, referring to FIG. 3 , the model loss of the image number recognition model is calculated according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the actual electric meter number ,include:
S301、根据所述真实电表区域和所述预测电表区域，计算所述图像数字识别模型的第一损失；S301. Calculate the first loss of the image digital recognition model according to the real meter area and the predicted meter area;
S302、根据所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的第二损失；S302. Calculate the second loss of the image number recognition model according to the predicted meter number and the real meter number;
S303、根据所述第一损失和所述第二损失，计算所述图像数字识别模型的模型损失。S303. Calculate the model loss of the image digital recognition model according to the first loss and the second loss.
一个可选实施例中，利用下述公式计算所述图像数字识别模型的第一损失：In an optional embodiment, the first loss of the image digit recognition model is calculated using the following formula:
L1＝mglogmp+(1-mg)log(1-mp)L1=mg logmp +(1-mg) log(1-mp )
其中，L1表示第一损失，mg表示预测电表区域的第g个像素点的像素值，mp表示真实电表区域的第p个像素点的像素值。Among them, L1 represents the first loss, mg represents the pixel value of the g-th pixel in the predicted meter area, and mp represents the pixel value of thep -th pixel in the real meter area.
一个可选实施例中，利用下述公式计算所述图像数字识别模型的第二损失：In an optional embodiment, the second loss of the image digit recognition model is calculated using the following formula:
L2＝|αp-αg|L2=|αp -αg |
其中，L2表示第二,损失，αg表示预测电表数字，αp表示真实电表数字。where L2 represents the second loss,αg represents the predicted meter number, andαp represents the real meter number.
一个可选实施例中，利用下述公式计算所述图像数字识别模型的模型损失：In an optional embodiment, the model loss of the image digit recognition model is calculated using the following formula:
L＝αL1+βL2L=αL1+βL2
其中，L表示模型损失，α表示第一损失的权重，β表示第二损失的权重。需要说明的是，本发明实施例中，所述图像数字识别模型最终的输出结果是电表数字，因此，所述β的值可以设置为远大于α，如β设置为0.8，α设置为0.2。Among them, L represents the model loss, α represents the weight of the first loss, and β represents the weight of the second loss. It should be noted that, in this embodiment of the present invention, the final output result of the image number recognition model is the meter number, therefore, the value of β can be set to be much larger than α, for example, β is set to 0.8, and α is set to 0.2.
基于所述模型损失的计算，可以作为所述图像数字识别模型是否具有较强识别能力的判断依据，从而可以判断所述图像数字识别模型是否需要继续训练The calculation based on the model loss can be used as a basis for judging whether the image digital recognition model has strong recognition ability, so as to determine whether the image digital recognition model needs to continue training
S4、判断所述模型损失是否满足预设条件。S4. Determine whether the model loss satisfies a preset condition.
本发明实施例通过判断所述模型损失是否满足预设条件，以识别所述图像数字识别模型是否需要继续训练，其中，所述预设条件可以设置为所述模型损失是否小于预设损失，即所述模型损失小于预设损失，则所述模型损失满足预设条件，所述模型损失不小于预设损失，则所述模型损失不满足预设条件，可选的，所述预设损失可以设置为0.1，也可以根据实际业务场景设置。In this embodiment of the present invention, it is determined whether the model loss satisfies a preset condition to identify whether the image digital recognition model needs to continue training. The preset condition may be set as whether the model loss is less than a preset loss, that is, If the model loss is less than the preset loss, the model loss satisfies the preset condition, and the model loss is not less than the preset loss, then the model loss does not meet the preset condition. Optionally, the preset loss can be It is set to 0.1, which can also be set according to actual business scenarios.
若所述模型损失不满足预设条件，执行S5、调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤。If the model loss does not meet the preset condition, perform S5, adjust the parameters of the image number recognition model, and return to performing the step of inputting the target dial image into the pre-built image number recognition model.
应该了解，在所述模型损失不满足预设条件时，表示所述图像数字识别模型并不具备较强的图像数字识别能力，因此本发明实施例通过调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤，以实现所述图像数字识别模型的继续训练，保障所述图像数字识别模型的识别能力。其中，所述图像数字识别模型的参数包括权重和偏置，所述图像数字识别模型的参数调整可以通过梯度下降算法实现，如随机梯度下降算法。It should be understood that when the model loss does not meet the preset conditions, it means that the image digital recognition model does not have strong image digital recognition capabilities. Return to execute the step of inputting the target dial image into the pre-built image digit recognition model, so as to realize the continuous training of the image digit recognition model and ensure the recognition ability of the image digit recognition model. Wherein, the parameters of the image digital recognition model include weights and biases, and the parameter adjustment of the image digital recognition model can be implemented by a gradient descent algorithm, such as a stochastic gradient descent algorithm.
若所述模型损失满足预设条件，执行S6、得到训练完成的图像数字识别模型。If the model loss satisfies the preset condition, perform S6 to obtain a trained image digital recognition model.
应该了解，在所述模型损失满足预设条件时，表示所述图像数字识别模型具备较强的图像数字识别能力，因此本发明直接生成训练完成的图像数字识别模型。It should be understood that when the model loss satisfies the preset conditions, it means that the image number recognition model has strong image number recognition capability, so the present invention directly generates a trained image number recognition model.
S7、利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。S7. Use the trained image digital recognition model to perform digital recognition on the dial image of the current electric meter to obtain the dial number of the current electric meter.
本发明实施例中，所述当前电表的表盘图像是指需要识别出电表数字的图像，本发明实施例通过将当前电表的表盘图像输入至所述训练完成的图像数字识别模型中，以智能化的输出所述所述当前电表的表盘数字，提高所述当前电表的表盘数字识别效率。In the embodiment of the present invention, the dial image of the current electric meter refers to an image that needs to identify the number of the electric meter. In the embodiment of the present invention, the dial image of the current electric meter is input into the image number recognition model after training, so as to realize intelligent The output of the dial number of the current electric meter improves the recognition efficiency of the dial number of the current electric meter.
可以看出，本发明实施例首先对历史电表的表盘图像进行预处理操作，得到目标表盘图像，可以减少后续模型的训练无用数据，提高模型训练的速度和质量，并标注目标表盘图像的真实电表区域及真实电表数字，可以作为后续模型训练结果的数据参照，监督模型的学习，提高模型的识别能力；其次，本发明实施例通过将目标表盘图像输入至预构建的图像数字识别模型中，以通过目标表盘图像对图像数字识别模型中的区域检测网络和文字识别网络进行训练，得到训练完成的图像数字识别模型，以减少后续人为参与电表数字的读取工作，提高后续电表的读取效率；进一步地，本发明实施例利用训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到当前电表的表盘数字，实现当前电表的数字智能读取。因此，本发明实施例提出的一种电表表盘数字的智能识别方法可以实现电表数字的自动识别方法，提高电表数字的读取效率。It can be seen that the embodiment of the present invention first performs a preprocessing operation on the dial image of the historical electricity meter to obtain the target dial image, which can reduce the useless data of subsequent model training, improve the speed and quality of model training, and mark the real electricity meter of the target dial image. The area and the real meter number can be used as a data reference for the subsequent model training results to supervise the learning of the model and improve the recognition ability of the model; The region detection network and text recognition network in the image digital recognition model are trained through the target dial image, and the trained image digital recognition model is obtained, so as to reduce the subsequent manual participation in the reading of the meter numbers and improve the reading efficiency of the subsequent meters; Further, the embodiment of the present invention uses the trained image digital recognition model to perform digital recognition on the dial image of the current electric meter, obtains the dial number of the current electric meter, and realizes the digital intelligent reading of the current electric meter. Therefore, the intelligent identification method for the numbers on the dial of the electric meter proposed by the embodiment of the present invention can realize the automatic identification method of the numbers of the electric meter, and improve the reading efficiency of the numbers of the electric meter.
如图4所示，是本发明电表表盘数字的智能识别装置的功能模块图。As shown in FIG. 4 , it is a functional block diagram of the intelligent identification device for the numbers on the dial of the electric meter of the present invention.
本发明所述电表表盘数字的智能识别装置400可以安装于电子设备中。根据实现的功能，所述电表表盘数字的智能识别装置可以包括图像预处理模块401模型训练模块402以及电表数字识别模块403。本发明所述模块也可以称之为单元，是指一种能够被电子设备处理器所执行，并且能够完成固定功能的一系列计算机程序段，其存储在电子设备的存储器中。The
在本实施例中，关于各模块/单元的功能如下：In this embodiment, the functions of each module/unit are as follows:
所述图像预处理模块401，用于采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字；The
所述模型训练模块402，用于将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字；The
所述模型训练模块402，还用于根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失；The
所述模型训练模块402，还用于判断所述模型损失是否满足预设条件；The
所述模型训练模块402，还用于在所述模型损失不满足所述预设条件时，调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤；The
所述模型训练模块402，还用于在所述模型损失满足所述预设条件时，得到训练完成的图像数字识别模型；The
所述电表数字识别模块403，用于利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。The electric meter
详细地，本发明实施例中所述电表表盘数字的智能识别装置400中的所述各模块在使用时采用与上述的图1至图3中所述的电表表盘数字的智能识别方法一样的技术手段，并能够产生相同的技术效果，这里不再赘述。In detail, the modules in the
如图5所示，是本发明实现电表表盘数字的智能识别方法的电子设备的结构示意图。As shown in FIG. 5 , it is a schematic structural diagram of the electronic device for realizing the intelligent identification method of the numbers on the dial of the electric meter according to the present invention.
所述电子设备可以包括处理器50、存储器51、通信总线52以及通信接口53，还可以包括存储在所述存储器51中并可在所述处理器50上运行的计算机程序，如电表表盘数字的智能识别程序。其中，所述处理器50在一些实施例中可以由集成电路组成，例如可以由单个封装的集成电路所组成，也可以是由多个相同功能或不同功能封装的集成电路所组成，包括一个或者多个中央处理器(Central Processing unit，CPU)、微处理器、数字处理芯片、图形处理器及各种控制芯片的组合等。所述处理器50是所述电子设备的控制核心(ControlUnit)，利用各种接口和线路连接整个电子设备的各个部件，通过运行或执行存储在所述存储器51内的程序或者模块(例如执行电表表盘数字的智能识别程序等)，以及调用存储在所述存储器51内的数据，以执行电子设备的各种功能和处理数据。The electronic device may include a
所述存储器51至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、移动硬盘、多媒体卡、卡型存储器(例如：SD或DX存储器等)、磁性存储器、磁盘、光盘等。所述存储器51在一些实施例中可以是电子设备的内部存储单元，例如该电子设备的移动硬盘。所述存储器51在另一些实施例中也可以是电子设备的外部存储设备，例如电子设备上配备的插接式移动硬盘、智能存储卡(Smart Media Card，SMC)、安全数字(Secure Digital，SD)卡、闪存卡(Flash Card)等。进一步地，所述存储器51还可以既包括电子设备的内部存储单元也包括外部存储设备。所述存储器51不仅可以用于存储安装于电子设备的应用软件及各类数据，例如电表表盘数字的智能识别程序的代码等，还可以用于暂时地存储已经输出或者将要输出的数据。The
所述通信总线52可以是外设部件互连标准(peripheral componentinterconnect，简称PCI)总线或扩展工业标准结构(extended industry standardarchitecture，简称EISA)总线等。该总线可以分为地址总线、数据总线、控制总线等。所述总线被设置为实现所述存储器51以及至少一个处理器50等之间的连接通信。The
所述通信接口53用于上述电子设备与其他设备之间的通信，包括网络接口和用户接口。可选地，所述网络接口可以包括有线接口和/或无线接口(如WI-FI接口、蓝牙接口等)，通常用于在该电子设备与其他电子设备之间建立通信连接。所述用户接口可以是显示器(Display)、输入单元(比如键盘(Keyboard))，可选地，所述用户接口还可以是标准的有线接口、无线接口。可选地，在一些实施例中，显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode，有机发光二极管)触摸器等。其中，显示器也可以适当的称为显示屏或显示单元，用于显示在电子设备中处理的信息以及用于显示可视化的用户界面。The
图5仅示出了具有部件的电子设备，本领域技术人员可以理解的是，图5示出的结构并不构成对所述电子设备的限定，可以包括比图示更少或者更多的部件，或者组合某些部件，或者不同的部件布置。FIG. 5 only shows an electronic device with components. Those skilled in the art can understand that the structure shown in FIG. 5 does not constitute a limitation on the electronic device, and may include fewer or more components than those shown in the drawings. , or a combination of certain components, or a different arrangement of components.
例如，尽管未示出，所述电子设备还可以包括给各个部件供电的电源(比如电池)，优选地，电源可以通过电源管理装置与所述至少一个处理器50逻辑相连，从而通过电源管理装置实现充电管理、放电管理、以及功耗管理等功能。电源还可以包括一个或一个以上的直流或交流电源、再充电装置、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。所述电子设备还可以包括多种传感器、蓝牙模块、Wi-Fi模块等，在此不再赘述。For example, although not shown, the electronic device may also include a power source (such as a battery) for powering the various components, preferably, the power source may be logically connected to the at least one
应该了解，所述实施例仅为说明之用，在专利发明范围上并不受此结构的限制。It should be understood that the embodiments are only used for illustration, and are not limited by this structure in the scope of the patented invention.
所述电子设备中的所述存储器51存储的电表表盘数字的智能识别程序是多个计算机程序的组合，在所述处理器50中运行时，可以实现：The intelligent identification program of the meter dial numbers stored in the
采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字；Collect the dial image of the historical electricity meter, perform a preprocessing operation on the dial image, obtain the target dial image, and mark the real electricity meter area and the real electricity meter number of the target dial image;
将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字；The target dial image is input into a pre-built image digital recognition model to detect the predicted meter area of the target dial image through the area detection network in the image digital recognition model, and use the image digital recognition model to detect the predicted meter area. The text recognition network identifies the predicted meter number in the predicted meter area;
根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失；Calculate the model loss of the image number recognition model according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the real electric meter number;
判断所述模型损失是否满足预设条件；Determine whether the model loss satisfies a preset condition;
若所述模型损失不满足所述预设条件，调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤；If the model loss does not meet the preset condition, adjust the parameters of the image number recognition model, and return to executing the step of inputting the target dial image into the pre-built image number recognition model;
若所述模型损失满足所述预设条件，得到训练完成的图像数字识别模型；If the model loss satisfies the preset condition, a trained image digital recognition model is obtained;
利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。Using the trained image number recognition model to perform digital recognition on the dial image of the current electric meter to obtain the dial number of the current electric meter.
具体地，所述处理器50对上述计算机程序的具体实现方法可参考图1对应实施例中相关步骤的描述，在此不赘述。Specifically, for the specific implementation method of the above-mentioned computer program by the
进一步地，所述电子设备集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个非易失性计算机可读取存储介质中。所述计算机可读存储介质可以是易失性的，也可以是非易失性的。例如，所述计算机可读介质可以包括：能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM，Read-Only Memory)。Further, if the modules/units integrated in the electronic device are implemented in the form of software functional units and sold or used as independent products, they may be stored in a non-volatile computer-readable storage medium. The computer-readable storage medium may be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a removable hard disk, a magnetic disk, an optical disc, a computer memory, a read-only memory (ROM, Read-Only). Memory).
本发明还提供一种计算机可读存储介质，所述可读存储介质存储有计算机程序，所述计算机程序在被电子设备的处理器所执行时，可以实现：The present invention also provides a computer-readable storage medium, where the readable storage medium stores a computer program, and when executed by a processor of an electronic device, the computer program can realize:
采集历史电表的表盘图像，对所述表盘图像进行预处理操作，得到目标表盘图像，并标注所述目标表盘图像的真实电表区域及真实电表数字；Collect the dial image of the historical electricity meter, perform a preprocessing operation on the dial image, obtain the target dial image, and mark the real electricity meter area and the real electricity meter number of the target dial image;
将所述目标表盘图像输入至预构建的图像数字识别模型中，以通过所述图像数字识别模型中的区域检测网络检测所述目标表盘图像的预测电表区域，并利用所述图像数字识别模型中的文字识别网络识别所述预测电表区域的预测电表数字；The target dial image is input into a pre-built image digital recognition model to detect the predicted meter area of the target dial image through the area detection network in the image digital recognition model, and use the image digital recognition model to detect the predicted meter area. The text recognition network identifies the predicted meter number in the predicted meter area;
根据所述真实电表区域和所述预测电表区域，及所述预测电表数字和所述真实电表数字，计算所述图像数字识别模型的模型损失；Calculate the model loss of the image number recognition model according to the real electric meter area and the predicted electric meter area, and the predicted electric meter number and the real electric meter number;
判断所述模型损失是否满足预设条件；Determine whether the model loss satisfies a preset condition;
若所述模型损失不满足所述预设条件，调整所述图像数字识别模型的参数，并返回执行所述将所述目标表盘图像输入至预构建的图像数字识别模型中的步骤；If the model loss does not meet the preset condition, adjust the parameters of the image number recognition model, and return to executing the step of inputting the target dial image into the pre-built image number recognition model;
若所述模型损失满足所述预设条件，得到训练完成的图像数字识别模型；If the model loss satisfies the preset condition, a trained image digital recognition model is obtained;
利用所述训练完成的图像数字识别模型对当前电表的表盘图像进行数字识别，得到所述当前电表的表盘数字。Using the trained image number recognition model to perform digital recognition on the dial image of the current electric meter to obtain the dial number of the current electric meter.
在本发明所提供的几个实施例中，应该理解到，所揭露的设备，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述模块的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there may be other division manners in actual implementation.
所述作为分离部件说明的模块可以是或者也可以不是物理上分开的，作为模块显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。The modules described as separate components may or may not be physically separated, and components shown as modules may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.
另外，在本发明各个实施例中的各功能模块可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能模块的形式实现。In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware, or can be implemented in the form of hardware plus software function modules.
对于本领域技术人员而言，显然本发明不限于上述示范性实施例的细节，而且在不背离本发明的精神或基本特征的情况下，能够以其他的具体形式实现本发明。It will be apparent to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, but that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics of the invention.
因此，无论从哪一点来看，均应将实施例看作是示范性的，而且是非限制性的，本发明的范围由所附权利要求而不是上述说明限定，因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括在本发明内。不应将权利要求中的任何附关联图标记视为限制所涉及的权利要求。Therefore, the embodiments are to be regarded in all respects as illustrative and not restrictive, and the scope of the invention is to be defined by the appended claims rather than the foregoing description, which are therefore intended to fall within the scope of the claims. All changes within the meaning and range of the equivalents of , are included in the present invention. Any reference signs in the claims shall not be construed as limiting the involved claim.
需要说明的是，在本文中，诸如“第一”和“第二”等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。It should be noted that, in this document, relational terms such as "first" and "second" etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply these There is no such actual relationship or sequence between entities or operations. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass a non-exclusive inclusion such that a process, method, article or device that includes a list of elements includes not only those elements, but also includes not explicitly listed or other elements inherent to such a process, method, article or apparatus. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in a process, method, article or apparatus that includes the element.
以上所述仅是本发明的具体实施方式，使本领域技术人员能够理解或实现本发明。对这些实施例的多种修改对本领域的技术人员来说将是显而易见的，本文中所定义的一般原理可以在不脱离本发明的精神或范围的情况下，在其它实施例中实现。因此，本发明将不会被限制于本文所示的这些实施例，而是要符合与本文所发明的原理和新颖特点相一致的最宽的范围。The above descriptions are only specific embodiments of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
Translated fromChinese
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210549315.6ACN114898367A (en) | 2022-05-20 | 2022-05-20 | Intelligent identification method, device, equipment and medium for electric meter dial numbers |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210549315.6ACN114898367A (en) | 2022-05-20 | 2022-05-20 | Intelligent identification method, device, equipment and medium for electric meter dial numbers |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114898367Atrue CN114898367A (en) | 2022-08-12 |
Family
ID=82723099
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210549315.6APendingCN114898367A (en) | 2022-05-20 | 2022-05-20 | Intelligent identification method, device, equipment and medium for electric meter dial numbers |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114898367A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119251849A (en)* | 2024-12-05 | 2025-01-03 | 江苏卡欧万泓电子有限公司 | Automatic meter reading method of electric energy meter based on computer vision |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875722A (en)* | 2017-12-27 | 2018-11-23 | 北京旷视科技有限公司 | Character recognition and recognition model training method, device and system, and storage medium |
| CN110427943A (en)* | 2019-08-27 | 2019-11-08 | 国家电网有限公司 | A kind of intelligent electric meter technique for partitioning based on R-CNN |
| CN112861648A (en)* | 2021-01-19 | 2021-05-28 | 平安科技（深圳）有限公司 | Character recognition method and device, electronic equipment and storage medium |
| CN113642582A (en)* | 2021-08-13 | 2021-11-12 | 中国联合网络通信集团有限公司 | Ammeter reading identification method and device, electronic equipment and storage medium |
| CN114241583A (en)* | 2022-01-19 | 2022-03-25 | 中国平安人寿保险股份有限公司 | Face recognition method and device, electronic equipment and storage medium |
| US20220129731A1 (en)* | 2021-05-27 | 2022-04-28 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and apparatus for training image recognition model, and method and apparatus for recognizing image |
- 2022
- 2022-05-20CNCN202210549315.6Apatent/CN114898367A/enactivePending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875722A (en)* | 2017-12-27 | 2018-11-23 | 北京旷视科技有限公司 | Character recognition and recognition model training method, device and system, and storage medium |
| CN110427943A (en)* | 2019-08-27 | 2019-11-08 | 国家电网有限公司 | A kind of intelligent electric meter technique for partitioning based on R-CNN |
| CN112861648A (en)* | 2021-01-19 | 2021-05-28 | 平安科技（深圳）有限公司 | Character recognition method and device, electronic equipment and storage medium |
| US20220129731A1 (en)* | 2021-05-27 | 2022-04-28 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and apparatus for training image recognition model, and method and apparatus for recognizing image |
| CN113642582A (en)* | 2021-08-13 | 2021-11-12 | 中国联合网络通信集团有限公司 | Ammeter reading identification method and device, electronic equipment and storage medium |
| CN114241583A (en)* | 2022-01-19 | 2022-03-25 | 中国平安人寿保险股份有限公司 | Face recognition method and device, electronic equipment and storage medium |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119251849A (en)* | 2024-12-05 | 2025-01-03 | 江苏卡欧万泓电子有限公司 | Automatic meter reading method of electric energy meter based on computer vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108593260B (en) | Optical cable line fault location and detection method and terminal equipment | |
| CN113781485B (en) | A kind of intelligent detection method, device, electronic device and medium for PCB defect category | |
| CN115469590B (en) | Low-power consumption control method, device and equipment for intelligent electric meter interface and storage medium | |
| CN111814743A (en) | Handwriting recognition method, device and computer-readable storage medium | |
| CN112149880A (en) | User scale prediction method, device, electronic equipment and storage medium | |
| CN116680580A (en) | Information matching method and device based on multi-mode training, electronic equipment and medium | |
| WO2023123926A1 (en) | Artificial intelligence task processing method and apparatus, electronic device, and readable storage medium | |
| CN116818090A (en) | Detection service intelligent analysis management system based on big data and cloud computing | |
| CN111651292A (en) | Data verification method, apparatus, electronic device and computer-readable storage medium | |
| CN114898367A (en) | Intelligent identification method, device, equipment and medium for electric meter dial numbers | |
| CN111339072B (en) | User behavior-based change value analysis method and device, electronic equipment and medium | |
| CN116434955A (en) | Method and device for evaluating employees' health status | |
| CN116580800A (en) | A method and device for constructing a reference interval | |
| CN111709346A (en) | Recognition and detection method of historical buildings based on deep learning and high-scoring images | |
| CN115577933A (en) | Intelligent park incubation enterprise information management system and method | |
| CN112561500B (en) | Salary data generation method, device, equipment and medium based on user data | |
| CN112686244B (en) | Automatic approval method, device, equipment and medium based on picture processing interface | |
| CN112113638B (en) | Water meter function self-checking device and method | |
| CN117195100A (en) | Prediction method, device and electronic equipment for current transformer error state | |
| CN117633680A (en) | Metering standard analysis method, device and equipment for direct-current electric energy meter and storage medium | |
| CN114936687A (en) | Data-driven urban daily water consumption prediction method | |
| CN114663877A (en) | Intelligent identification method, device, equipment and medium for steel plate blank number | |
| CN114742801A (en) | A kind of B-ultrasound image intelligent recognition method and device based on PSO-SVM algorithm | |
| WO2022141838A1 (en) | Model confidence analysis method and apparatus, electronic device and computer storage medium | |
| CN116990744B (en) | Electric energy meter detection method, device, equipment and medium |
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 |