CN111666930A - Meter reading method, device, system and storage medium - Google Patents

Meter reading method, device, system and storage medium Download PDF

Info

Publication number
CN111666930A
CN111666930A CN201910171358.3A CN201910171358A CN111666930A CN 111666930 A CN111666930 A CN 111666930A CN 201910171358 A CN201910171358 A CN 201910171358A CN 111666930 A CN111666930 A CN 111666930A
Authority
CN
China
Prior art keywords
neural network
instrument
convolutional neural
image
meter
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201910171358.3A
Other languages
Chinese (zh)
Inventor
黄河
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Haiqing Zhiying Technology Co ltd
Original Assignee
Shenzhen Haiqing Zhiying Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Haiqing Zhiying Technology Co ltd filed Critical Shenzhen Haiqing Zhiying Technology Co ltd
Priority to CN201910171358.3A priority Critical patent/CN111666930A/en
Publication of CN111666930A publication Critical patent/CN111666930A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/10—Image acquisition
    • G06V10/12—Details of acquisition arrangements; Constructional details thereof
    • G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
    • G06V10/143—Sensing or illuminating at different wavelengths
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Arrangements For Transmission Of Measured Signals (AREA)

Abstract

The application relates to a meter reading method, a device, a system and a storage medium. The method comprises the following steps: acquiring an indication image of the instrument acquired by the optical sensor; transmitting the indication image to a convolutional neural network dedicated chip; processing the indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of a target instrument; and outputting the reading value of the target instrument. By adopting the method, the meter data can be automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the meter data is improved.

Description

Meter reading method, device, system and storage medium
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a method, an apparatus, a system, and a storage medium for reading a meter.
Background
With the development of scientific technology, more and more traditional pointer type instruments are replaced by digital instruments, but the digital instruments are easily interfered by electromagnetic fields to cause the errors of readings, so that a large number of pointer type instruments are still used in certain specific occasions.
However, most of the conventional pointer instrument reading methods are manually performed, but errors are generated due to interference of human factors in the manual reading methods.
Disclosure of Invention
In view of the above, it is necessary to provide a meter reading method, apparatus, system and storage medium for solving the above technical problems.
A meter reading method, the method comprising:
acquiring an indication image of the instrument acquired by the optical sensor;
transmitting the indication image to a convolutional neural network dedicated chip;
processing the indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of a target instrument;
and outputting the reading value of the target instrument.
In one embodiment, the training step of the convolutional neural network model comprises:
acquiring a distance between the optical sensor and the meter;
acquiring the instrument type of the instrument;
constructing a sample set of indication images according to the distance and the instrument type;
training a convolutional neural network model based on the set of indicative image samples.
In one embodiment, the training step of the convolutional neural network model further comprises:
acquiring a meter image of the meter, the meter image being acquired by one of the optical sensor and a backup optical sensor disposed near the optical sensor;
the obtaining the distance between the optical sensor and the meter comprises:
carrying out image analysis on the instrument image to obtain the distance between the optical sensor and the instrument;
the obtaining of the instrument type of the instrument includes:
and carrying out image recognition on the instrument image to obtain the instrument type of the instrument.
In one embodiment, said processing said indication image through a convolutional neural network model on said convolutional neural network dedicated chip to obtain a target meter reading value comprises:
preprocessing the indication images to ensure that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after preprocessing;
and processing the preprocessed indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indication image by the convolutional neural network model.
In one embodiment, the outputting the target meter reading value comprises:
and sending the reading value of the target instrument to target equipment in a wireless transmission mode.
A meter reading system, the system comprising an optical sensor, a processor, and a convolutional neural network specific chip; the optical sensor and the convolution neural network special chip are respectively connected with the processor;
the optical sensor is used for sending the acquired indication image of the instrument to the processor;
the processor is used for transmitting the instrument indication image to a convolution neural network special chip;
the convolution neural network special chip is used for processing the instrument indication image through a convolution neural network model on the convolution neural network special chip to obtain a target instrument reading value;
the processor is also used for outputting the reading value of the target instrument.
In one embodiment, the system further comprises a memory connected with the processor, wherein the memory stores a convolutional neural network model for loading to the convolutional neural network special chip for operation; the convolutional neural network model is obtained by constructing an indication image sample set according to the distance and the instrument type after the distance between the optical sensor and the instrument type of the instrument are obtained, and training based on the indication image sample set.
In one embodiment, the distance is obtained by image analysis of a meter image of the meter, the type of the meter is obtained by image recognition of the meter image, and the meter image is acquired by one of the optical sensor and a backup optical sensor disposed near the optical sensor.
In one embodiment, the processor is further configured to pre-process the indication images such that different indication images acquired under different image acquisition conditions have a uniform image acquisition condition after being pre-processed;
the convolutional neural network special chip is also used for processing the preprocessed indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indication image by the convolutional neural network model.
In one embodiment, the system further comprises a wireless communication unit connected to the processor;
the processor is used for outputting the reading value of the target meter to the wireless communication unit;
the wireless communication unit is used for sending the target instrument reading value to target equipment in a wireless transmission mode.
A meter reading apparatus, the apparatus comprising:
the indicating image acquisition module is used for acquiring an indicating image of the instrument acquired by the optical sensor;
the indication image transmission module is used for transmitting the indication image to the convolution neural network special chip;
the convolutional neural network model processing module is used for processing the indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument;
and the instrument reading value output module is used for outputting the target instrument reading value.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indication image by the convolutional neural network model.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned meter reading method.
According to the instrument reading method, the device, the system and the storage medium, the indication image of the instrument is acquired by the optical sensor arranged outside the instrument, the acquired indication image is transmitted to the convolution neural network special chip, and the indication image is processed by the convolution neural network model on the convolution neural network special chip to obtain the reading value of the target instrument, so that the instrument data is automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the instrument data is improved.
Drawings
FIG. 1 is a diagram of an embodiment of a meter reading method;
FIG. 2 is a flow diagram illustrating a meter reading method in one embodiment;
FIG. 3 is a schematic view of a meter reading system in one embodiment;
FIG. 4 is a block diagram of a meter reading device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The meter reading method provided by the application can be applied to the application environment shown in fig. 1. The application environment includes a meter reading system that includes an optical sensor 102, a processor 104, and a convolutional neural network specific chip 106. The optical sensor 102 and the convolution neural network dedicated chip are respectively connected with the processor. Wherein, the optical sensor 102 can be one or more than one. The processor 104 may be a single-core or multi-core processor. The convolutional neural network dedicated chip 106 has a chip structure adapted to the convolutional neural network model, and may be a Field Programmable Gate Array (FPGA) chip or an Application Specific Integrated Circuit (ASIC) chip.
In one embodiment, as shown in fig. 2, a meter reading method is provided, and the present embodiment is described by taking the example that the method is applied to the meter reading system in fig. 1, and includes the following steps:
step S202, an indication image of the meter collected by the optical sensor is acquired.
The optical sensor is a sensor for carrying out photosensitive imaging on a target object according to an optical principle. Meters are instruments that indicate the results of a measurement, including pressure meters, flow meters, thermometers, and analytical instruments. The meter may be an electronic type or a mechanical type.
The meter may be a digital or a graphical indicating type meter. The numerical indication type is a meter type in which a measurement result of a meter is directly presented in a numerical form, and the graphic indication type is a meter type in which a measurement result is graphically represented. A meter of the graphic indication type, such as a meter of the scale type, such as a meter indicating the scale by means of a pointer or a fluid.
The indication image is an image including a measurement result indicated by the meter, and may specifically be a partial image of a part of the meter for indicating the measurement result. When the meter is of the digital indication type, the corresponding indication image may be an image of a display area for displaying the measurement result in digital form in the meter. When the meter is a scale type meter, the corresponding indication image may be an image of a scale area indicated by the fluid in the meter.
Specifically, at least one optical sensor is arranged outside the meter to be read, an indication image of the meter is collected through the optical sensor, the collected indication image of the corresponding meter is sent to the processor, and the collected indication image of the corresponding meter can also be stored in the optical sensor. The processor can receive the indication image of the corresponding meter sent by the optical sensor, and can also directly acquire the indication image of the corresponding meter from the optical sensor.
In one embodiment, the optical sensor may specifically be a device that collects an indication image of the meter in real time and transmits the collected image to the processor in real time or stores the collected image in the optical sensor in real time; it is also possible to periodically collect an indication image of the meter and periodically transmit the collected indication image to the processor or periodically store the acquired indication image in the optical sensor.
And step S204, transmitting the indication image to a convolution neural network special chip.
Specifically, the processor transmits the acquired indication image of the instrument to a special chip of the convolutional neural network. The convolution neural network special chip is provided with a chip structure matched with the convolution neural network model, and the chip structure is used for optimizing the processing logic of the convolution neural network model when the indication image of the instrument is processed through the convolution neural network model. The convolutional neural network model is one of neural network models, and is a deep learning model having convolutional layers.
And step S206, processing the indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a reading value of the target instrument.
The target meter reading value is a specific value indicated by the image. Specifically, a chip dedicated to the convolutional neural network receives an indication image of the instrument transmitted through the processor, and a convolutional neural network model is run on the chip dedicated to the convolutional neural network. And processing the received indication image through a convolutional neural network model running on a convolutional neural network special chip so as to obtain a reading value of the target instrument.
And step S208, outputting the reading value of the target meter.
The output mode may be a wireless transmission mode or a wired transmission mode, such as a radio frequency transmission mode, an NFC (near field communication) transmission mode, a bluetooth transmission mode, or a wireless network transmission mode. Specifically, the processor outputs the target meter reading value, and specifically, the obtained target meter reading value can be sent out through the communication unit, can be visually displayed through the display, and can also be output in a voice mode through the loudspeaker. The communication unit may be a wireless communication unit or a wired communication unit.
In one embodiment, step S208 includes: and sending the reading value of the target instrument to the target equipment in a wireless transmission mode. Specifically, the processor can output the target meter reading value to the wireless communication unit, and the wireless communication unit sends the target meter reading value to the target device in a wireless transmission mode. The target device may be a pre-configured device, may be any device nearby, or may be a device that is currently wirelessly connected to the meter reading system.
In the embodiment, the indication image of the instrument is acquired by the optical sensor arranged outside the instrument, the acquired indication image is transmitted to the special convolutional neural network chip, and the indication image is processed by the convolutional neural network model on the special convolutional neural network chip to obtain the reading value of the target instrument, so that the instrument data is automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the instrument data is improved.
In one particular embodiment, as shown in FIG. 3, a meter reading system is provided. The meter reading system comprises an optical sensor, a processor, a convolutional neural network special chip, a memory, a communication unit and a power supply circuit. The optical sensor, the memory, the convolution neural network special chip and the communication unit are respectively connected with the processor; the power circuit provides electric quantity for the whole system, and the power supply can be a battery, a solar power supply or direct current.
Specifically, the memory stores a computer program and a convolutional neural network model, and when the meter reading system runs, the computer program is loaded into the processor to be executed, and the convolutional neural network model is loaded onto a chip special for the convolutional neural network to run. The optical sensor collects an indication image of the instrument and transmits the collected indication image to the processor; the processor transmits the indication image to a convolution neural network special chip, and the indication image of the instrument is processed through a convolution neural network model on the convolution neural network special chip so as to obtain a reading value of the target instrument; and the processor outputs the reading value of the target instrument by adopting a pre-configured output mode.
In one embodiment, the training step of the convolutional neural network model comprises: acquiring the distance between the optical sensor and the instrument; acquiring the instrument type of the instrument; constructing an indication image sample set according to the distance and the instrument type; a convolutional neural network model is trained based on a set of indicative image samples.
Wherein, the instrument types are divided according to the types of the instruments. When the meter is a digital indication type meter, the corresponding meter type is a meter type in which the measurement result of the meter is directly presented in a digital form; when the meter is a meter of a graphic indication type, the corresponding meter type is a meter type graphically representing the measurement result. The set of indication image samples comprises a plurality of indication image samples, each of which may be constructed according to the type of meter and the distance between the optical sensor and the meter.
Specifically, a training device may be preset, and the training device is configured to train the convolutional neural network model and store the trained convolutional neural network model in a memory in the meter reading system. Acquiring the distance between the optical sensor and the instrument type of the instrument through training equipment; constructing a corresponding indication image sample according to the acquired distance and the instrument type; further, performing space geometric transformation on the indication image samples with different reading values to form an indication image sample set; wherein the spatial geometrical transformation comprises a scaling transformation, a rotation transformation and a shape transformation. And training the convolutional neural network model running on a convolutional neural network special chip by the training equipment based on the constructed indicating image sample set so as to obtain the trained convolutional neural network model with the corresponding distance and instrument type.
In one embodiment, the training process of the convolutional neural network model may also be performed by using a processor in the meter reading system, and the trained convolutional neural network model is stored in a memory for loading by the convolutional neural network chip.
In one embodiment, the training device may obtain a distance between the optical sensor and the meter, a meter type of the meter, and an orientation of the optical sensor relative to the meter, wherein the orientation includes a direction of the optical sensor relative to the meter and an angle of the optical sensor relative to the meter. And further, constructing a corresponding indication image sample according to the acquired distance, the acquired instrument type and the acquired direction.
In one embodiment, the training device may obtain the illumination, the distance between the optical sensor and the meter, and the meter type of the meter in different collection environments; further, corresponding indication image samples are constructed according to the obtained illumination, the distance and the instrument type, and indication image samples in different illumination environments are obtained. Specifically, acquiring the illumination in different acquisition environments may include: and acquiring illumination corresponding to a plurality of different time nodes in one day.
In one embodiment, the training device acquires noise, distance between the optical sensor and the meter, and the meter type of the meter under different acquisition environments; and constructing a corresponding indication image sample according to the acquired noise, distance and instrument type to obtain an indication image sample in a noise environment. Specifically, acquiring the noise under different acquisition environments may include: the training equipment randomly generates a group of noise points, so that the noise points represent the noise in different acquisition environments.
In one embodiment, the training device acquires the instrument type of the instrument, and constructs an indication image sample of a corresponding type according to the acquired instrument type; further, based on the constructed indication image sample, the size of the component for indicating the measurement result is adjusted, thereby generating indication image samples of different sizes. The dimensions of the component for indicating the measurement result characterize the distance between the optical sensor and the meter; if the size of the component is large, the optical sensor is close to the instrument; if the size of the component is small, it means that the distance between the optical sensor and the meter is long.
In one embodiment, the training device may obtain the meter type of the meter; selecting an indicating image corresponding to the instrument type from the existing indicating images according to the instrument type as an indicating image sample; further, the size of the part for indicating the measurement result is adjusted based on the selected indication image sample, thereby generating indication image samples of different sizes.
In the embodiment, the indicating image sample set is constructed according to one or more combinations of the distance between the optical sensor and the meter, the type of the meter, illumination and noise, so that various ways of constructing the indicating image sample set are provided, and the indicating image sample set is enriched, so that the constructed indicating image sample set can cover images generated under various environments.
In one embodiment, the training step of the convolutional neural network model further comprises: acquiring an instrument image of an instrument, wherein the instrument image is acquired by one of an optical sensor and a standby optical sensor arranged near the optical sensor; acquiring the distance between the optical sensor and the meter includes: carrying out image analysis on the instrument image to obtain the distance between the optical sensor and the instrument; the instrument type of the access instrument includes: and carrying out image recognition on the instrument image to obtain the instrument type of the instrument.
Wherein the backup optical sensor is another optical sensor disposed in the vicinity of the optical sensor. The back up sensor may be one or more than one. The meter image is an image including the meter.
Specifically, at least one standby optical sensor is arranged outside the instrument to be read, and an instrument image of the instrument is collected through the standby optical sensor; the instrument image of the instrument can also be collected by an optical sensor arranged outside the instrument. The collected instrument images of the corresponding instruments are transmitted to the processor, and the collected instrument images of the corresponding instruments can also be stored in the corresponding collecting units, wherein the collecting units comprise standby optical sensors and optical sensors. The processor receives an instrument image of a corresponding instrument, and performs image analysis on the received instrument image to obtain the size of a part used for indicating a measurement result in the instrument image; further, the distance between the optical sensor and the meter is determined according to the size of the component. The processor receives an instrument image of a corresponding instrument, and performs image recognition on the received instrument image to obtain the shape of the corresponding instrument; the type of the meter is further determined based on the shape of the meter.
In one embodiment, the processor may also directly acquire an instrument image of a corresponding instrument from the corresponding acquisition unit, and perform image analysis on the acquired instrument image to obtain a distance between the optical sensor and the instrument; and carrying out image recognition on the acquired instrument image to obtain the instrument type.
In one embodiment, the processor may perform image recognition on the received instrument image to obtain a product model of the corresponding instrument; further, the type of the instrument is determined according to the product model of the instrument.
In this embodiment, a distance between the optical sensor and the meter is obtained by acquiring a meter image of the meter and performing image analysis on the acquired meter image; carrying out image recognition on the acquired instrument image to obtain the instrument type of the instrument; a distance parameter and an instrument type parameter are provided for indicating construction of a sample set of images.
In one embodiment, processing the indication image through a convolutional neural network model on a chip dedicated to the convolutional neural network to obtain the target meter reading value comprises: preprocessing the indication images to ensure that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after preprocessing; and (4) processing the preprocessed indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a reading value of the target instrument.
The image acquisition condition refers to the environment of the acquisition unit when acquiring the indication image, and the image acquisition condition comprises illumination during image acquisition and the direction of the acquisition unit relative to the target instrument.
Specifically, the processor receives an indication image of the instrument collected by the optical sensor, and preprocesses the received indication image, so that different indication images collected under different image collection conditions have uniform image collection conditions after being preprocessed. The processor can also directly acquire the indication image of the corresponding instrument from the optical sensor and preprocess the acquired indication image. For example, under different lighting conditions, the indicating images of the corresponding meters are collected through the optical sensor, and the collected indicating images are transmitted to the processor; the processor preprocesses the indication images collected under different illumination conditions, so that the preprocessed indication images have uniform illumination.
Further, the processor transmits the preprocessed indication image to the special convolutional neural network chip, and processes the received preprocessed indication image through the convolutional neural network model running on the special convolutional neural network chip so as to obtain the reading value of the target instrument. Or the preprocessed indication image can be stored in the processor, and the special chip for the convolutional neural network directly acquires the preprocessed indication image from the processor.
In the embodiment, the collected indication images are preprocessed, so that different indication images collected under different image collection conditions have uniform image collection conditions after being preprocessed, and errors caused by different image collection conditions on reading values of the target instrument are avoided; furthermore, the preprocessed indication image is processed through the convolutional neural network model on the convolutional neural network special chip, so that the processing speed of the convolutional neural network special chip is improved, and the reading value of the target instrument can be quickly obtained.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indicative image by the convolutional neural network model.
Specifically, the convolutional neural network dedicated chip has a chip structure adapted to the convolutional neural network model, and when the indication image is processed by the convolutional neural network model on the convolutional neural network dedicated chip, the chip structure can optimize the processing logic of the convolutional neural network model, convert serial processing into parallel processing, and thereby improve the rate of data processing.
In this embodiment, the chip dedicated for the convolutional neural network has a chip structure adapted to the convolutional neural network model, and when the convolutional neural network model processes the indication image, the chip structure can optimize the processing logic of the convolutional neural network model, and improve the processing rate.
It should be understood that, although the steps in the flowchart of fig. 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In one embodiment, as shown in FIG. 3, a meter-reading system includes an optical sensor, a processor, and a convolutional neural network specific chip. The optical sensor and the convolution neural network special chip are respectively connected with the processor. The optical sensor is used for sending the acquired indication image of the instrument to the processor; the processor is used for transmitting the instrument indication image to the convolution neural network special chip; the convolution neural network special chip is used for processing the instrument indication image through a convolution neural network model on the convolution neural network special chip to obtain a target instrument reading value; the processor is also used for outputting the reading value of the target instrument.
In one embodiment, the system further comprises a memory connected with the processor, wherein the memory stores a convolutional neural network model for loading to a special chip of the convolutional neural network for operation; the convolutional neural network model is obtained by constructing an indication image sample set according to the distance and the instrument type after the distance between the optical sensor and the instrument type of the instrument are obtained, and training the indication image sample set.
In one embodiment, the distance is obtained by image analysis of a meter image of the meter, the type of the meter is obtained by image recognition of the meter image, and the meter image is acquired by one of the optical sensor and a backup optical sensor disposed near the optical sensor.
In one embodiment, the processor is further configured to pre-process the indication images such that different indication images acquired under different image acquisition conditions have a uniform image acquisition condition after being pre-processed. The convolution neural network special chip is also used for processing the preprocessed indication image through a convolution neural network model on the convolution neural network special chip to obtain a reading value of the target instrument.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indicative image by the convolutional neural network model.
In one embodiment, the system further comprises a wireless communication unit coupled to the processor. The processor is also used for outputting the reading value of the target meter to the wireless communication unit. The wireless communication unit is used for sending the reading value of the target instrument to the target equipment in a wireless transmission mode.
In the embodiment, the indication image of the instrument is acquired by the optical sensor arranged outside the instrument, the acquired indication image is transmitted to the convolution neural network special chip, and the indication image is processed by the convolution neural network model on the convolution neural network special chip to obtain the reading value of the target instrument, so that the instrument data is automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the instrument data is improved.
In one embodiment, as shown in FIG. 4, there is provided a meter-reading apparatus 400 comprising: an indication image acquisition module 402, an indication image transmission module 404, a convolutional neural network model processing module 406, and an instrument reading value output module 408, wherein:
and an indication image acquisition module 402, configured to acquire an indication image of the meter acquired by the optical sensor.
And an indication image transmission module 404, configured to transmit the indication image to the convolutional neural network dedicated chip.
And the convolutional neural network model processing module 406 is configured to process the indication image through a convolutional neural network model on a chip dedicated to the convolutional neural network, so as to obtain a reading value of the target instrument.
And an instrument reading value output module 408, configured to output a target instrument reading value.
In one embodiment, the convolutional neural network model processing module is further configured to: acquiring the distance between the optical sensor and the instrument; acquiring the instrument type of the instrument; constructing an indication image sample set according to the distance and the instrument type; a convolutional neural network model is trained based on a set of indicative image samples.
In one embodiment, the convolutional neural network model processing module is further configured to: acquiring an instrument image of an instrument, wherein the instrument image is acquired by one of an optical sensor and a standby optical sensor arranged near the optical sensor; carrying out image analysis on the instrument image to obtain the distance between the optical sensor and the instrument; and carrying out image recognition on the instrument image to obtain the instrument type of the instrument.
In one embodiment, the convolutional neural network model processing module is further configured to: preprocessing the indication images to ensure that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after preprocessing; and (4) processing the preprocessed indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a reading value of the target instrument.
In one embodiment, the convolutional neural network specific chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize the processing logic of the convolutional neural network model when processing the indicative image by the convolutional neural network model.
In one embodiment, the meter reading value output module is further configured to send the target meter reading value to the target device through a wireless transmission manner.
In the embodiment, the indication image of the instrument is acquired by the optical sensor arranged outside the instrument, the acquired indication image is transmitted to the special convolutional neural network chip, and the indication image is processed by the convolutional neural network model on the special convolutional neural network chip to obtain the reading value of the target instrument, so that the instrument data is automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the instrument data is improved.
For specific limitations of the meter reading device, reference may be made to the limitations of the meter reading method above, and further description thereof is omitted here. The various modules in the meter reading apparatus described above may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the system, and can also be stored in a memory in the system in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of: acquiring an indication image of the instrument acquired by the optical sensor; transmitting the indication image to a convolution neural network special chip; processing the indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a reading value of the target instrument; and outputting the reading value of the target instrument.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring the distance between the optical sensor and the instrument; acquiring the instrument type of the instrument; constructing an indication image sample set according to the distance and the instrument type; a convolutional neural network model is trained based on a set of indicative image samples.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring an instrument image of an instrument, wherein the instrument image is acquired by one of an optical sensor and a standby optical sensor arranged near the optical sensor; carrying out image analysis on the instrument image to obtain the distance between the optical sensor and the instrument; and carrying out image recognition on the instrument image to obtain the instrument type of the instrument.
In one embodiment, the computer program when executed by the processor further performs the steps of: processing the indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a target instrument reading value, wherein the step of processing the indication image through the convolutional neural network model on the convolutional neural network special chip comprises the following steps: preprocessing the indication images to ensure that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after preprocessing; and (4) processing the preprocessed indication image through a convolutional neural network model on a convolutional neural network special chip to obtain a reading value of the target instrument.
In one embodiment, the computer program when executed by the processor further performs the steps of: the convolution neural network special chip is provided with a chip structure matched with the convolution neural network model, and the chip structure is used for optimizing the processing logic of the convolution neural network model when the indication image is processed through the convolution neural network model.
In one embodiment, the computer program when executed by the processor further performs the steps of: outputting the target meter reading value includes: and sending the reading value of the target instrument to the target equipment in a wireless transmission mode.
In the embodiment, the indication image of the instrument is acquired by the optical sensor arranged outside the instrument, the acquired indication image is transmitted to the convolution neural network special chip, and the indication image is processed by the convolution neural network model on the convolution neural network special chip to obtain the reading value of the target instrument, so that the instrument data is automatically read with low power consumption, the interference of human factors is eliminated, and the accuracy of reading the instrument data is improved.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (15)

1. A meter reading method, the method comprising:
acquiring an indication image of the instrument acquired by the optical sensor;
transmitting the indication image to a convolutional neural network dedicated chip;
processing the indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of a target instrument;
and outputting the reading value of the target instrument.
2. The method of claim 1, wherein the step of training the convolutional neural network model comprises:
acquiring a distance between the optical sensor and the meter;
acquiring the instrument type of the instrument;
constructing a sample set of indication images according to the distance and the instrument type;
training a convolutional neural network model based on the set of indicative image samples.
3. The method of claim 2, wherein the step of training the convolutional neural network model further comprises:
acquiring a meter image of the meter, the meter image being acquired by one of the optical sensor and a backup optical sensor disposed near the optical sensor;
the obtaining the distance between the optical sensor and the meter comprises:
carrying out image analysis on the instrument image to obtain the distance between the optical sensor and the instrument;
the obtaining of the instrument type of the instrument includes:
and carrying out image recognition on the instrument image to obtain the instrument type of the instrument.
4. The method of claim 1, wherein said processing the indication image through a convolutional neural network model on the convolutional neural network dedicated chip to obtain a target meter reading value comprises:
preprocessing the indication images to ensure that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after preprocessing;
and processing the preprocessed indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument.
5. The method of claim 1, wherein the convolutional neural network dedicated chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize processing logic of the convolutional neural network model when processing the indicative image through the convolutional neural network model.
6. The method of any one of claims 1 to 5, wherein the outputting the target meter reading value comprises:
and sending the reading value of the target instrument to target equipment in a wireless transmission mode.
7. A meter reading system, comprising an optical sensor, a processor, and a convolutional neural network specific chip; the optical sensor and the convolution neural network special chip are respectively connected with the processor;
the optical sensor is used for sending the acquired indication image of the instrument to the processor;
the processor is used for transmitting the instrument indication image to a convolution neural network special chip;
the convolution neural network special chip is used for processing the instrument indication image through a convolution neural network model on the convolution neural network special chip to obtain a target instrument reading value;
the processor is also used for outputting the reading value of the target instrument.
8. The system of claim 7, further comprising a memory coupled to the processor, the memory storing a convolutional neural network model for loading into the convolutional neural network specific chip for operation; the convolutional neural network model is obtained by constructing an indication image sample set according to the distance and the instrument type after the distance between the optical sensor and the instrument type of the instrument are obtained, and training based on the indication image sample set.
9. The system of claim 8, wherein the distance is an image analysis of a meter image of the meter, wherein the meter type is an image recognition of the meter image, and wherein the meter image is captured by one of the optical sensor and a backup optical sensor disposed proximate to the optical sensor.
10. The system of claim 7, wherein the processor is further configured to pre-process the indication images such that different indication images acquired under different image acquisition conditions have uniform image acquisition conditions after being pre-processed;
the convolutional neural network special chip is also used for processing the preprocessed indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument.
11. The system of claim 7, wherein the convolutional neural network dedicated chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize processing logic of the convolutional neural network model when processing the indicative image through the convolutional neural network model.
12. The system of any one of claims 7 to 11, further comprising a wireless communication unit connected to the processor;
the processor is used for outputting the reading value of the target meter to the wireless communication unit;
the wireless communication unit is used for sending the target instrument reading value to target equipment in a wireless transmission mode.
13. A meter reading apparatus, the apparatus comprising:
the indicating image acquisition module is used for acquiring an indicating image of the instrument acquired by the optical sensor;
the indication image transmission module is used for transmitting the indication image to the convolution neural network special chip;
the convolutional neural network model processing module is used for processing the indication image through a convolutional neural network model on the convolutional neural network special chip to obtain a reading value of the target instrument;
and the instrument reading value output module is used for outputting the target instrument reading value.
14. The apparatus of claim 13, wherein the convolutional neural network dedicated chip has a chip structure adapted to the convolutional neural network model, the chip structure being used to optimize processing logic of the convolutional neural network model when processing the indication image through the convolutional neural network model.
15. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
CN201910171358.3A 2019-03-07 2019-03-07 Meter reading method, device, system and storage medium Pending CN111666930A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910171358.3A CN111666930A (en) 2019-03-07 2019-03-07 Meter reading method, device, system and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910171358.3A CN111666930A (en) 2019-03-07 2019-03-07 Meter reading method, device, system and storage medium

Publications (1)

Publication Number Publication Date
CN111666930A true CN111666930A (en) 2020-09-15

Family

ID=72382015

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910171358.3A Pending CN111666930A (en) 2019-03-07 2019-03-07 Meter reading method, device, system and storage medium

Country Status (1)

Country Link
CN (1) CN111666930A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013136295A1 (en) * 2012-03-15 2013-09-19 Northstar Telemetrics, S. L. Method for automatically reading a utility meter, retrofittable meter reader and automatic meter reading system using the same
CN105809179A (en) * 2014-12-31 2016-07-27 中国科学院深圳先进技术研究院 Pointer type instrument reading recognition method and device
CN107229930A (en) * 2017-04-28 2017-10-03 北京化工大学 A kind of pointer instrument numerical value intelligent identification Method and device
CN108830271A (en) * 2018-06-13 2018-11-16 深圳市云识科技有限公司 A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks
CN109190565A (en) * 2018-09-06 2019-01-11 苏州立升净水科技有限公司 Instrumented data acquisition method, device and filtration system
CN109214378A (en) * 2018-08-16 2019-01-15 新智数字科技有限公司 A kind of method and system integrally identifying metering meter reading based on neural network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013136295A1 (en) * 2012-03-15 2013-09-19 Northstar Telemetrics, S. L. Method for automatically reading a utility meter, retrofittable meter reader and automatic meter reading system using the same
CN105809179A (en) * 2014-12-31 2016-07-27 中国科学院深圳先进技术研究院 Pointer type instrument reading recognition method and device
CN107229930A (en) * 2017-04-28 2017-10-03 北京化工大学 A kind of pointer instrument numerical value intelligent identification Method and device
CN108830271A (en) * 2018-06-13 2018-11-16 深圳市云识科技有限公司 A kind of digital displaying meter Recognition of Reading method based on convolutional neural networks
CN109214378A (en) * 2018-08-16 2019-01-15 新智数字科技有限公司 A kind of method and system integrally identifying metering meter reading based on neural network
CN109190565A (en) * 2018-09-06 2019-01-11 苏州立升净水科技有限公司 Instrumented data acquisition method, device and filtration system

Similar Documents

Publication Publication Date Title
EP2860596B1 (en) Methods and apparatus relating to measurement instruments
JP7183425B2 (en) Machine learning system and data fusion to optimize deployment conditions for detecting corrosion under insulation
US10088411B2 (en) Method and device for performing colorimetric analysis of a test fluid to evaluate physiological parameters
US20140028819A1 (en) Measuring endoscope apparatus and program
CN107076571A (en) Method, equipment and system for initializing meter reading device
EP3767551A1 (en) Inspection system, image recognition system, recognition system, discriminator generation system, and learning data generation device
US10533937B1 (en) Cloud-based machine learning system and data fusion for the prediction and detection of corrosion under insulation
ES2527737T3 (en) Method to visualize the temperature field of a biological subject
CN109633529B (en) Detection equipment, method and device for positioning accuracy of positioning system
CN112985615A (en) Body temperature monitoring method and device
KR101584604B1 (en) Disease decoding system by analyzing the quantity of motion of livestock
CN111666930A (en) Meter reading method, device, system and storage medium
JP6649244B2 (en) Automatic inspection system and automatic inspection method
CN112990107A (en) Hyperspectral remote sensing image underwater target detection method and device and computer equipment
CN115436570B (en) Method and device for remotely sensing and monitoring carbon dioxide concentration based on multivariate data
KR101636410B1 (en) Method, server, terminal, system, and recording medium for diagnosing thermal insulation property of building
CN113343554B (en) Arch dam underwater damage identification method, terminal equipment and storage medium
US20230148924A1 (en) Information processing device, control method, and storage medium
KR101757258B1 (en) Mobile device with character recognition for the visiting metering and method of visiting metering using thereof
CN110609038B (en) Structural damage identification method and system based on unmanned aerial vehicle image
US20150145688A1 (en) Pipeline Sensor System and Method
CN206312350U (en) Integrated environment data intelligence harvester
US20230237073A1 (en) Taxonomy-based diagnostic system and method, and readable storage medium
CN105388784B (en) The data collection of adaptability and state-driven
CN111222364B (en) Meter reading method and meter reading device

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