CN110569769A - image recognition method and device, computer equipment and storage medium - Google Patents
image recognition method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110569769A CN110569769A CN201910806890.8A CN201910806890A CN110569769A CN 110569769 A CN110569769 A CN 110569769A CN 201910806890 A CN201910806890 A CN 201910806890A CN 110569769 A CN110569769 A CN 110569769A
- Authority
- CN
- China
- Prior art keywords
- image
- neural network
- identification
- network model
- training
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The application relates to an image recognition method, an image recognition device, a computer device and a storage medium. The method comprises the following steps: collecting an image to be detected; selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image; and identifying the region-of-interest image according to a preset identification model, and acquiring identification information in the region-of-interest image. By adopting the method, the image processing technology can be utilized, the image to be detected is acquired according to the acquired image, and the acquired image is input into the preset convolutional neural network model and the preset identification model to acquire the identification image in the acquired image, so that the labor cost is reduced, and the efficiency and the accuracy are improved.
Description
Technical Field
the present application relates to the field of image recognition technologies, and in particular, to an image recognition method, an image recognition apparatus, a computer device, and a storage medium.
Background
with the development of social economy and the improvement of the living standard of urban residents, vehicles become basic requirements of people's life, China is a large population country and also a large number of motor vehicles, and with the rapid development of the economy of China, the number of motor vehicles is rapidly increased.
At present, registration and registration of license plates on vehicles involve a plurality of department units and links, and various procedures need to be provided and handled. Specifically, when a vehicle is being checked, a vehicle-handing check is performed, including taking a picture of the vehicle, checking the engine topology of the vehicle, and the like. After the vehicle is delivered and checked, various procedures of the vehicle, such as vehicle purchase tax completion certification, traffic compulsory insurance policy, vehicle purchase invoice, vehicle quality qualification certificate, identity card, and the like, are required to be carried out for card loading.
in the traditional vehicle license plate management, the information is input mainly by manpower, the data of the vehicle is manually input, and the data is input into a database, so that the data can be safely and efficiently managed, and the follow-up procedures are simplified. However, in the current process of registering the license plate of the vehicle, the data are mainly manually input by a registering worker, and in the process of manual input, the processing efficiency is low and the accuracy is not high.
disclosure of Invention
In view of the above, it is necessary to provide an image recognition method, an image recognition apparatus, a computer device, and a storage medium, which can achieve high processing efficiency and high accuracy.
in a first aspect, a method of image recognition, the method comprising:
collecting an image to be detected;
Selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
And identifying the region-of-interest image according to a preset identification model, and acquiring identification information in the region-of-interest image.
In one embodiment, the selecting, from the acquired image, an image of a region of interest in the image according to the convolutional neural network model includes: the convolutional neural network model comprises a first convolutional neural network model and a second convolutional neural network model;
Inputting the image to be detected into a first convolution neural network model to obtain a coarse positioning image;
And inputting the coarse positioning image into a second convolutional neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image.
in one embodiment, the selecting the region of interest image in the image from the acquired image according to the convolutional neural network model comprises:
The region of interest comprises a first identification region and/or a second identification region;
Establishing a first neural network, training a neural network model by taking a training image containing an identification image as a training set to obtain a first convolution neural network model, wherein the input of the first convolution neural network model is the training image, and the output of the first convolution neural network model is a coarse positioning image in which the first identification area is located in the training image;
and establishing a second neural network, training a neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, wherein the input of the second convolutional neural network model is the training image, and the output of the second convolutional neural network model is the fine positioning image in which the second identification area is positioned in the coarse positioning image.
in one embodiment, the identifying the region-of-interest image according to a preset identification model, before acquiring the identification image in the region-of-interest image, includes:
and establishing a recognition model, and training the recognition model by taking the fine positioning image as a training set to obtain the recognition model, wherein the input of the recognition model is the fine positioning image, and the output of the recognition model is identification information.
In one embodiment, the establishing a recognition model, training the recognition model by using the fine positioning image as a training set, and obtaining the recognition model further includes;
Preprocessing the fine positioning image to generate a sample image, wherein the preprocessing is used for adjusting image parameters of the sample image;
And training a recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model.
In one embodiment, the preprocessing the training image to adjust the image parameters of the training image includes:
performing one or more of blurring, modifying contrast, modifying brightness, or image enhancement processing on the training image;
The image parameter includes at least one of a resolution size, a brightness, a contrast, a rotation angle, or a color.
In one embodiment, the method further comprises:
Checking whether the identification information is correct;
if the identification information is correct, outputting the identification information;
And if the identification information is wrong, re-identifying the acquired image.
In a second aspect, an image recognition apparatus, the apparatus comprising:
the image acquisition module is used for acquiring an image to be detected;
The region dividing module is used for selecting an interested region image in the image from the acquired image according to the convolutional neural network model, wherein the interested region image comprises an identification image;
and the image identification module is used for identifying the interested area image according to a preset identification model and acquiring an identification image in the interested area image.
In a third aspect, a computer device comprises a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
collecting an image to be detected;
selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
and identifying the region-of-interest image according to a preset identification model, and acquiring an identification image in the region-of-interest image.
in a fourth aspect, a computer readable storage medium having stored thereon a computer program which when executed by a processor implements the steps of:
Collecting an image to be detected;
selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
And identifying the region-of-interest image according to a preset identification model, and acquiring identification information in the region-of-interest image.
according to the image recognition method, the image recognition device, the computer equipment and the storage medium, the image to be detected is collected by using an image processing technology, and is input into the preset convolutional neural network model and the preset recognition model to obtain the identification information in the collected image, so that the labor cost is reduced, and the efficiency and the accuracy are improved.
drawings
FIG. 1 is a flow diagram illustrating an image recognition method according to one embodiment;
FIG. 2 is a flow chart illustrating an image recognition method according to another embodiment;
FIG. 3 is a block diagram showing the structure of an image recognition apparatus according to an embodiment;
FIG. 4 is a diagram illustrating an internal structure of a computer device according to an 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.
In recent years, with the emergence of big data, the strong computing power of computer hardware and the rapid development of related technologies of neural networks, deep learning makes an important breakthrough in the field of artificial intelligence, and has made great success in various fields such as natural language processing, voice recognition, computer vision, image and video analysis, and the like. Recently, deep learning is also increasingly applied in medical image processing and analysis, such as: image recognition and detection, image segmentation, image registration, image generation, image denoising and the like. The main difference between deep learning and the traditional pattern recognition method is that the deep learning can automatically learn from big data to obtain effective features, but not manually design the features, the manually designed features mainly depend on the prior knowledge of designers, the advantages of the big data are difficult to utilize, and the good features can effectively improve the performance of the pattern recognition system. The deep learning model not only can utilize context information in an image, but also implicitly adds the shape prior of an object through learning existing data in a high-dimensional data conversion process. The deep learning model comprises an unsupervised learning model and a supervised learning model, and a deep Convolutional Neural Network (CNN) is the most representative deep learning model with supervised learning, and is mainly applied to the image field.
OCR (Optical Character Recognition) refers to a process in which an electronic device (e.g., a scanner or a digital camera) checks a Character printed on paper, determines its shape by detecting dark and light patterns, and then translates the shape into a computer text by a Character Recognition method; the method is characterized in that characters in a paper document are converted into an image file with a black-white dot matrix in an optical mode aiming at print characters, and the characters in the image are converted into a text format through recognition software for further editing and processing by word processing software. How to debug or use auxiliary information to improve Recognition accuracy is the most important issue of OCR, and the term of ICR (Intelligent Character Recognition) is generated accordingly. The main indicators for measuring the performance of an OCR system are: the rejection rate, the false recognition rate, the recognition speed, the user interface friendliness, the product stability, the usability, the feasibility and the like.
the image identification method can be applied to computer equipment. The computer device may be a personal computer, a notebook computer, a smart phone, a tablet computer, a portable wearable device, a server, or a server cluster composed of a plurality of servers.
In one embodiment, as shown in fig. 1, an image recognition method is provided, which is described by taking the example that the method is applied to the terminal in fig. 1, and includes the following steps:
Step 102, collecting an image to be detected.
the image comprises an identification image, wherein the identification image is text information, color information, depth information or other identification information; it should be understood here that the text information comprises one or more of letters, numbers or symbols.
In one embodiment, the identification image in the captured image includes a vehicle GPS device number, a vehicle VIN number, a license plate number, an engine number, or other vehicle information.
Specifically, an image carrying an identification image is acquired through acquisition equipment, and the acquired image is preprocessed to adjust image parameters of the acquired image. The preprocessing mode is one or more of cropping, rotating or brightness conversion processing on the acquired image. The preprocessing process can also comprise normalization, namely, the acquired image is normalized according to a normalization mode adopted in a convolutional neural network model training stage, so that the convergence speed of the convolutional neural network model is increased.
it should be understood here that the acquisition device is a camera, recorder, video camera, or the like having an image acquisition function.
And 104, selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image.
the region of interest comprises a first identification region and a second identification region, and the first identification region comprises the second identification region.
It should be understood here that the first identification area is an image area containing color information, may also be an image area containing text information, and may also be an image area containing depth information; or an image area simultaneously containing one or more of text information, color information or depth information; the second identification area is also arranged identically, which is not repeated herein.
specifically, an interested area image in the image is selected from the collected image according to the convolutional neural network model, and the collected image is subjected to coarse positioning image recognition to obtain a coarse positioning image in which a first identification area in the training image is located; and performing fine positioning image recognition on the coarse positioning image to obtain a fine positioning image where the second identification area is located in the coarse positioning image, and more accurately obtaining the area where the identification image is located through multiple times of image recognition. Carrying out coarse positioning image recognition on the collected image to obtain an image of a first identification area where the identification image is located, namely a coarse positioning image; and performing fine positioning image recognition on the coarse positioning image to obtain an image of a second identification area where the identification image is located, namely the fine positioning image. The calculation force required by coarse positioning image identification is small, the identification speed is high, the rough area is quickly positioned, the calculation force required by fine positioning image identification is large, the identification speed is slower than that of the coarse positioning image identification, the identification speed is improved by reducing the identification area, and the positioning difficulty can be reduced, the accuracy rate is improved, and the identification speed of the convolutional neural network model is accelerated by adopting a mode of combining the coarse positioning image identification and the fine positioning image identification. The rough positioning image recognition needs to obtain an interested area image in the collected image, the area range is large, the characteristics are obvious, the difficulty is low, the accuracy is relatively high, the obtained image also comprises a part except the identification image, and the edge of the obtained interested area image is horizontal relative to the edge of the collected image; the fine positioning image recognition needs to obtain an identification area image along the identification image, the area of the character occupying image is small, the positioning difficulty is high, the edge of the obtained image is likely to incline relative to the edge of the collected image, and the inclination angle needs to be calculated.
Before selecting the interested region image in the image according to the convolution neural network model from the acquired image, the method comprises the following steps: establishing a neural network, training a neural network model by taking a training image containing the identification image as a training set to obtain a convolutional neural network model, wherein the input of the convolutional neural network model is the training image, and the output of the convolutional neural network model is an interested region image where the identification image is located in the training image.
The convolutional neural network model can adopt a depth convolutional neural network, a full convolutional neural network, a U-Net neural network, a V-Net neural network, a lightweight convolutional neural network (MobileNet), a resnet-50 convolutional neural network or other types of neural network models based on patch, and only needs to be suitable for positioning and segmenting the image.
specifically, the convolutional neural network model includes a first convolutional neural network model and a second convolutional neural network model. Inputting an image to be detected into a first convolution neural network model to obtain a coarse positioning image; and inputting the coarse positioning image into a second convolution neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image. More specifically, a first neural network is established, a training image containing an identification image is used as a training set to train a neural network model, a first convolution neural network model is obtained, the input of the first convolution neural network model is the training image, and the output of the first convolution neural network model is a coarse positioning image where a first identification area in the training image is located; and establishing a second neural network, training the neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, inputting the second convolutional neural network model as a training image, and outputting the training image and the fine positioning image in which the second identification area in the coarse positioning image is positioned. In one embodiment, the convolutional neural network model is a first convolutional neural network model, only the first neural network is established, a training image containing an identification image is used as a training set to train the neural network model, the first convolutional neural network model is obtained, the input of the first convolutional neural network model is the training image, the output of the first convolutional neural network model is a coarse positioning image where a first identification area in the training image is located, and the recognition model recognizes the coarse positioning image. In another embodiment, the convolutional neural network model is a second convolutional neural network model, a second neural network is established, the neural network model is trained by taking a training image containing the identification image as a training set to obtain the second convolutional neural network model, the input of the second convolutional neural network model is the training image, the output of the second convolutional neural network model is a fine positioning image where a second identification area in the coarse positioning image is located, and the recognition model recognizes the fine positioning image. The preferred mode is to determine the area where the identification image is located by combining coarse positioning and fine positioning.
the method also comprises the following steps: inputting an image to be detected into a first convolution neural network model to obtain a coarse positioning image; and inputting the coarse positioning image into a second convolution neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image. Specifically, inputting an acquired image into a first convolution neural network model, roughly positioning an approximate area where an identification image in the acquired image is located by the first convolution neural network model, wherein the identification image accounts for 10% -30% of the area, and obtaining a roughly positioned image; and inputting the coarse positioning image into a second convolutional neural network model, accurately positioning an accurate region where the identification image is located in the coarse positioning image by the second convolutional neural network model, and obtaining a fine positioning image by the identification image accounting for 60% -90% of the region.
And 106, identifying the region-of-interest image according to a preset identification model, and acquiring identification information in the region-of-interest image.
The identification information is text information, and according to different use conditions, the identification information is also depth information or other data information.
Here, it should be understood that the identification information is a letter, a number, a chinese character, or the like when it is text information.
in one embodiment, the identification information is a vehicle GPS device number, vehicle VIN number, license plate number, engine number, or the like.
wherein the recognition model comprises a DenseNet dense convolutional network.
the method comprises the following steps: and establishing a recognition model, training the recognition model by taking the fine positioning image as a training set to obtain the recognition model, inputting the recognition model into the fine positioning image, and outputting the recognition model into identification information. Specifically, preprocessing the fine positioning image to generate a sample image, wherein the preprocessing is used for adjusting image parameters of the sample image; and training the recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model, and increasing the training intensity of the recognition model by complicating the sample image. Furthermore, the image parameters of the sample image are adjusted by preprocessing the sample image, the adjusted sample image and the fine positioning image are used as a training set to train the recognition model, the image parameters of the sample image are changed by image processing, the training intensity of the recognition model is increased, and the generalization ability of the trained recognition model is improved, wherein the generalization ability (generalization ability) refers to the recognition ability of a machine learning algorithm on unseen samples. The image processing mode is one or more of blurring, contrast modification, brightness modification or image enhancement processing on the training image. In one embodiment, the training image is one or more of blurred, modified contrast, modified brightness, or image enhancement. Wherein the image parameter includes at least one of resolution size, brightness, contrast, rotation angle or color.
the method also comprises the following steps: checking whether the identification information is correct; if the identification information is correct, outputting the identification information; and if the identification information is wrong, re-acquiring the image and identifying the acquired image.
Specifically, if the identification image is the VIN code, whether the identification information is correct is checked according to a checking algorithm; if the identification information is correct, outputting the identification information; and if the identification information is wrong, re-acquiring the image and identifying the acquired image. The checking algorithm is as follows: the VIN code is obtained by multiplying the weighted value of the first bit by the corresponding value of the code number, and calculating the remainder of the sum of the product values of all 17 bits divided by 11, i.e. the ninth bit check value. Wherein, the weight of the content is: VIN code "corresponding value" of each digit:
Weight of position VIN code from 1 st bit to 17 th bit "weight":
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 |
| 8 | 7 | 6 | 5 | 4 | 3 | 2 | 10 | 0 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 |
Vehicle identification code: the ninth bit 9 of the UU6JA69691D713820 is a check bit, and the verification mode is as follows:
4 × 8+4 × 7+6 × 6+1 × 5+1 × 4+6 × 3+9 × 2+6 × 10+1 × 9+4 × 8+7 × 7+1 × 6+3 × 5+8 × 4+2 × 3+0 × 0 is 350, 350 is divided by 11 to obtain 31, and the remainder 9 is the check code, and if the values of the check code and the check bit are the same, the VIN code correctly outputs the identification information, and if the values of the check code and the check bit are not the same, the VIN code incorrectly re-identifies the acquired image.
In one embodiment, an image to be detected is collected, the image to be detected is input into a first convolution neural network model to obtain a coarse positioning image, the coarse positioning image is input into a second convolution neural network model to obtain a fine positioning image, and the fine positioning image is used as an interested region image; and identifying the image of the region of interest according to a preset identification model, and acquiring identification information in the image of the region of interest. And identifying the identification information in the acquired image by the image to be detected through the first convolutional neural network model, the second convolutional neural network model and the identification model. Wherein, the recognition function of the recognition model adopts OCR technology.
according to the image recognition method, the image processing technology is utilized, the image to be detected is collected and input into the preset convolutional neural network model and the preset recognition model, and the identification information in the collected image is obtained, so that the labor cost is reduced, and the efficiency and the accuracy are improved.
In one embodiment, as shown in fig. 2, fig. 2 is a flow chart illustrating an image recognition method in another embodiment. In this embodiment, the identification information is a GPS device and a VIN code.
And preprocessing the acquired image carrying the identification image, wherein the preprocessing mode comprises rotation, cutting, brightness conversion, noise reduction or other image processing modes.
and inputting the preprocessed acquired image into a convolutional neural network model to obtain an image of the region of interest. Specifically, inputting the preprocessed acquired image into a first convolution neural network model to perform rough positioning of a candidate area where a VIN code and a GPS equipment number are located, and obtaining a rough positioning image; and inputting the coarse positioning image into a second convolutional neural network model to perform fine positioning VIN code and GPS equipment number character area to obtain a fine positioning image. In this embodiment, the first convolutional neural network model is a lightweight convolutional neural network, and the second convolutional neural network model is a resnet-50 convolutional neural network. Because the difficulty is low in roughly positioning the key area of the GPS equipment number and the key area of the VIN code, the lightweight convolutional neural network is adopted, so that high recall rate can be obtained, the processing speed is high, and the detection can be completed within dozens of milliseconds. The first convolution neural network model is designed to be 416 x 3 resolution images, an image with 13 x 512 resolution is obtained through 10 convolution layers, and a loss design of a YOLO model is added, so that the extraction of the area where the VIN code and the GPS equipment number are located is realized, the interference of other invalid information can be eliminated, the size of the processed image of fine positioning is reduced, and the calculation amount of the fine positioning is greatly reduced. The input size of the second convolutional neural network model is designed to be 320 × 320, the identification image in the coarse positioning image is finely positioned, the identification image in the coarse positioning image is extracted through the resnet-50 convolutional neural network, the resnet-50 convolutional neural network can extract deep features, and the resnet-50 convolutional neural network is not easy to cause unconvergence. And (3) carrying out scene image text detection (advanced EAST) on the extracted identification image, detecting the boundary of the area where the identification image is located, and extracting the image of the area along the boundary as a fine positioning image, so that the problems of inaccurate character edge positioning and small character line spacing can be solved to a certain extent.
and inputting the fine positioning image into a recognition model, and recognizing the identification image in the fine positioning image by the recognition model. In this embodiment, the input resolution of the recognition model is set to 512 × 64 × 1, features of the fine positioning image are extracted by using a DenseNet dense convolution network, the extracted image with the feature set size of 1 × 35 × 512 is processed by using a softmax function and a CTC function, the extracted features are converted into finally output characters, and identification information, namely, a VIN code and a GPS device number, is acquired.
Checking the VIN code through a checking algorithm, and if the VIN code is checked to be correct, outputting the VIN code; and if the VIN code is checked to be wrong, acquiring the image again and identifying the acquired image. It is understood that the VIN code may also be corrected manually.
It should be understood that although the various steps in the flow charts of fig. 1-2 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 some of the steps in fig. 1-2 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 performing the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.
in one embodiment, as shown in fig. 3, there is provided an image recognition apparatus including: an image acquisition module 210, a region division module 220, and an image recognition module 230, wherein:
An image acquisition module 210, configured to acquire an image to be detected;
The region dividing module 220 is configured to select an image of a region of interest in the image from the acquired image according to the convolutional neural network model, where the image of the region of interest includes the identification image.
The image recognition module 230 is configured to recognize the region-of-interest image according to a preset recognition model, and acquire an identification image in the region-of-interest image.
The region dividing module 220 is further configured to establish a neural network, train the neural network model by using a training image including the identification image as a training set, obtain a convolutional neural network model, input the convolutional neural network model as the training image, and output the training image as an image of a region of interest where the identification image is located in the training image.
The region dividing module 220 is further configured to establish a first neural network, train the neural network model by using a training image including the identification image as a training set, obtain a first convolution neural network model, input the first convolution neural network model as a training image, and output the training image as a coarse positioning image in which the first identification region is located in the training image; and establishing a second neural network, training the neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, inputting the second convolutional neural network model as a training image, and outputting the training image and the fine positioning image in which the second identification area in the coarse positioning image is positioned.
the region dividing module 220 is further configured to input the image to be detected into the first convolutional neural network model to obtain a coarse positioning image; and inputting the coarse positioning image into a second convolution neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image.
the image recognition module 230 is further configured to establish a recognition model, train the recognition model using the fine positioning image as a training set, obtain the recognition model, and output the recognition model as an identification image with the input of the fine positioning image.
the image recognition module 230 is further configured to perform preprocessing on the fine positioning image to generate a sample image, where the preprocessing is used to adjust image parameters of the sample image; and training the recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model. Performing one or more of blurring, modifying contrast, modifying brightness, or image enhancement on the training image; the image parameter includes at least one of a resolution size, brightness, contrast, rotation angle, or color.
The image recognition device also comprises a checking module used for checking whether the identification information is correct; if the identification information is correct, outputting the identification information; and if the identification information is wrong, re-acquiring the image and identifying the acquired image.
for specific limitations of the image recognition device, reference may be made to the above limitations of the image recognition method, which are not described herein again. The modules in the image recognition device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, the computer device is a server, and the internal structure diagram of the computer device can be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store image recognition data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an image recognition method.
Those skilled in the art will appreciate that the architecture shown in fig. 4 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
Collecting an image to be detected;
Selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
and identifying the image of the region of interest according to a preset identification model, and acquiring identification information in the image of the region of interest.
in one embodiment, the processor, when executing the computer program, further performs the steps of: establishing a first neural network, training a neural network model by taking a training image containing an identification image as a training set to obtain a first convolution neural network model, inputting the first convolution neural network model as the training image, and outputting the training image as a coarse positioning image of a first identification area in the training image; and establishing a second neural network, training the neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, inputting the second convolutional neural network model as a training image, and outputting the training image and the fine positioning image in which the second identification area in the coarse positioning image is positioned.
in one embodiment, the processor, when executing the computer program, further performs the steps of: inputting an image to be detected into a first convolution neural network model to obtain a coarse positioning image; and inputting the coarse positioning image into a second convolution neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image.
in one embodiment, the processor, when executing the computer program, further performs the steps of: and establishing a recognition model, training the recognition model by taking the fine positioning image as a training set to obtain the recognition model, inputting the recognition model into the fine positioning image, and outputting the recognition model into identification information.
In one embodiment, the processor, when executing the computer program, further performs the steps of: preprocessing the fine positioning image to generate a sample image, wherein the preprocessing is used for adjusting image parameters of the sample image; and training the recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: performing one or more of blurring, modifying contrast, modifying brightness, or image enhancement on the training image; the image parameter includes at least one of a resolution size, brightness, contrast, rotation angle, or color.
In one embodiment, the processor, when executing the computer program, further performs the steps of: checking whether the identification information is correct; if the identification information is correct, outputting the identification information; and if the identification information is wrong, re-acquiring the image and identifying the acquired image.
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:
Collecting an image to be detected;
selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
And identifying the image of the region of interest according to a preset identification model, and acquiring identification information in the image of the region of interest.
in one embodiment, the computer program when executed by the processor further performs the steps of: establishing a first neural network, training a neural network model by taking a training image containing an identification image as a training set to obtain a first convolution neural network model, inputting the first convolution neural network model as the training image, and outputting the training image as a coarse positioning image of a first identification area in the training image; and establishing a second neural network, training the neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, inputting the second convolutional neural network model as a training image, and outputting the training image and the fine positioning image in which the second identification area in the coarse positioning image is positioned.
in one embodiment, the computer program when executed by the processor further performs the steps of: inputting an image to be detected into a first convolution neural network model to obtain a coarse positioning image; and inputting the coarse positioning image into a second convolution neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image.
In one embodiment, the computer program when executed by the processor further performs the steps of: and establishing a recognition model, training the recognition model by taking the fine positioning image as a training set to obtain the recognition model, inputting the recognition model into the fine positioning image, and outputting the recognition model into identification information.
in one embodiment, the computer program when executed by the processor further performs the steps of: preprocessing the fine positioning image to generate a sample image, wherein the preprocessing is used for adjusting image parameters of the sample image; and training the recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model.
In one embodiment, the computer program when executed by the processor further performs the steps of: performing one or more of blurring, modifying contrast, modifying brightness, or image enhancement on the training image; the image parameter includes at least one of a resolution size, brightness, contrast, rotation angle, or color.
In one embodiment, the computer program when executed by the processor further performs the steps of: checking whether the identification information is correct; if the identification information is correct, outputting the identification information; and if the identification information is wrong, re-acquiring the image and identifying the acquired image.
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 (Rambus) direct RAM (RDRAM), direct memory 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 (10)
1. An image recognition method, characterized in that the method comprises:
Collecting an image to be detected;
Selecting an interested area image in the image from the acquired image according to the convolutional neural network model, wherein the interested area image comprises an identification image;
And identifying the region-of-interest image according to a preset identification model, and acquiring identification information in the region-of-interest image.
2. The method of claim 1, wherein selecting the image of the region of interest in the image from the acquired image according to the convolutional neural network model comprises: the convolutional neural network model comprises a first convolutional neural network model and a second convolutional neural network model;
inputting the image to be detected into a first convolution neural network model to obtain a coarse positioning image;
and inputting the coarse positioning image into a second convolutional neural network model to obtain a fine positioning image, and taking the fine positioning image as an interested region image.
3. the method of claim 1 or 2, wherein said selecting the region of interest image in the image from the acquired image according to the convolutional neural network model comprises:
The region of interest comprises a first identification region and/or a second identification region;
Establishing a first neural network, training a neural network model by taking a training image containing an identification image as a training set to obtain a first convolution neural network model, wherein the input of the first convolution neural network model is the training image, and the output of the first convolution neural network model is a coarse positioning image in which the first identification area is located in the training image;
And establishing a second neural network, training a neural network model by taking the coarse positioning image as a training set to obtain a second convolutional neural network model, wherein the input of the second convolutional neural network model is the training image, and the output of the second convolutional neural network model is the fine positioning image in which the second identification area is positioned in the coarse positioning image.
4. The method according to claim 2, wherein the identifying the region-of-interest image according to a preset identification model, and the obtaining of the identification information in the region-of-interest image comprises:
and establishing a recognition model, and training the recognition model by taking the fine positioning image as a training set to obtain the recognition model, wherein the input of the recognition model is the fine positioning image, and the output of the recognition model is identification information.
5. The method of claim 4, wherein the building of the recognition model, the training of the recognition model using the fine positioning image as a training set, and the obtaining of the recognition model further comprises;
Preprocessing the fine positioning image to generate a sample image, wherein the preprocessing is used for adjusting image parameters of the sample image;
And training a recognition model by taking the fine positioning image and the sample image as a training set to obtain the recognition model.
6. The method of claim 5, wherein the pre-processing the training image to adjust image parameters of the training image comprises:
performing one or more of blurring, modifying contrast, modifying brightness, or image enhancement processing on the training image;
The image parameter includes at least one of a resolution size, a brightness, a contrast, a rotation angle, or a color.
7. The method of claim 1, further comprising:
Checking whether the identification information is correct;
If the identification information is correct, outputting the identification information;
And if the identification information is wrong, re-acquiring the image and identifying the acquired image.
8. An image recognition apparatus, characterized in that the apparatus comprises:
The image acquisition module is used for acquiring an image to be detected;
The region dividing module is used for selecting an interested region image in the image from the acquired image according to the convolutional neural network model, wherein the interested region image comprises an identification image;
and the image identification module is used for identifying the interested area image according to a preset identification model and acquiring the identification information in the interested area image.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program.
10. 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 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910806890.8A CN110569769A (en) | 2019-08-29 | 2019-08-29 | image recognition method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910806890.8A CN110569769A (en) | 2019-08-29 | 2019-08-29 | image recognition method and device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110569769A true CN110569769A (en) | 2019-12-13 |
Family
ID=68776769
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910806890.8A Pending CN110569769A (en) | 2019-08-29 | 2019-08-29 | image recognition method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110569769A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401312A (en) * | 2020-04-10 | 2020-07-10 | 深圳新致软件有限公司 | PDF drawing character recognition method, system and equipment |
| CN111489347A (en) * | 2020-04-15 | 2020-08-04 | 深圳前海环融联易信息科技服务有限公司 | Business license picture quality detection method and device, computer equipment and storage medium |
| CN111568195A (en) * | 2020-02-29 | 2020-08-25 | 佛山市云米电器科技有限公司 | Brewed beverage identification method, device and computer-readable storage medium |
| CN111640088A (en) * | 2020-04-22 | 2020-09-08 | 深圳拓邦股份有限公司 | Electronic component polarity detection method and system based on deep learning and electronic device |
| CN111860027A (en) * | 2020-06-11 | 2020-10-30 | 贝壳技术有限公司 | Two-dimensional code identification method and device |
| CN111950468A (en) * | 2020-08-14 | 2020-11-17 | 广东工业大学 | Traffic condition detection method and system based on convolutional neural network and storage medium |
| CN111950550A (en) * | 2020-08-13 | 2020-11-17 | 王宗尧 | Vehicle frame number identification system based on deep convolutional neural network |
| CN112017165A (en) * | 2020-08-19 | 2020-12-01 | 上海美沃精密仪器股份有限公司 | Lacrimal river height detection method based on deep learning |
| CN112164057A (en) * | 2020-10-09 | 2021-01-01 | 珠海格力电器股份有限公司 | Qualified label detection method, storage medium and electronic equipment |
| CN112419299A (en) * | 2020-12-04 | 2021-02-26 | 中冶建筑研究总院（深圳）有限公司 | Bolt loss detection method, device, equipment and storage medium |
| CN112927197A (en) * | 2021-02-19 | 2021-06-08 | 中冶建筑研究总院（深圳）有限公司 | Air conditioner external unit support corrosion detection method, device, equipment and storage medium |
| CN113012189A (en) * | 2021-03-31 | 2021-06-22 | 影石创新科技股份有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN113139553A (en) * | 2020-01-16 | 2021-07-20 | 中国科学院国家空间科学中心 | U-net-based method and system for extracting aurora ovum form of ultraviolet aurora image |
| WO2021143088A1 (en) * | 2020-01-19 | 2021-07-22 | 深圳壹账通智能科技有限公司 | Synchronous check method and apparatus for multiple certificate types, and computer device and storage medium |
| CN113609900A (en) * | 2021-06-25 | 2021-11-05 | 南京信息工程大学 | Local generation face positioning method and device, computer equipment and storage medium |
| CN117765533A (en) * | 2024-02-22 | 2024-03-26 | 天津医科大学第二医院 | image processing method and system for oral mucosa cancer prediction |
| CN117765533B (en) * | 2024-02-22 | 2024-04-26 | 天津医科大学第二医院 | Image processing method and system for oral mucosa cancer prediction |
-
2019
- 2019-08-29 CN CN201910806890.8A patent/CN110569769A/en active Pending
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113139553A (en) * | 2020-01-16 | 2021-07-20 | 中国科学院国家空间科学中心 | U-net-based method and system for extracting aurora ovum form of ultraviolet aurora image |
| WO2021143088A1 (en) * | 2020-01-19 | 2021-07-22 | 深圳壹账通智能科技有限公司 | Synchronous check method and apparatus for multiple certificate types, and computer device and storage medium |
| CN111568195A (en) * | 2020-02-29 | 2020-08-25 | 佛山市云米电器科技有限公司 | Brewed beverage identification method, device and computer-readable storage medium |
| CN111401312A (en) * | 2020-04-10 | 2020-07-10 | 深圳新致软件有限公司 | PDF drawing character recognition method, system and equipment |
| CN111401312B (en) * | 2020-04-10 | 2024-04-26 | 深圳新致软件有限公司 | PDF drawing text recognition method, system and equipment |
| CN111489347A (en) * | 2020-04-15 | 2020-08-04 | 深圳前海环融联易信息科技服务有限公司 | Business license picture quality detection method and device, computer equipment and storage medium |
| CN111489347B (en) * | 2020-04-15 | 2023-10-31 | 深圳前海环融联易信息科技服务有限公司 | Business license picture quality detection method and device, computer equipment and storage medium |
| CN111640088A (en) * | 2020-04-22 | 2020-09-08 | 深圳拓邦股份有限公司 | Electronic component polarity detection method and system based on deep learning and electronic device |
| CN111640088B (en) * | 2020-04-22 | 2023-12-01 | 深圳拓邦股份有限公司 | Electronic element polarity detection method and system based on deep learning and electronic device |
| CN111860027A (en) * | 2020-06-11 | 2020-10-30 | 贝壳技术有限公司 | Two-dimensional code identification method and device |
| CN111950550A (en) * | 2020-08-13 | 2020-11-17 | 王宗尧 | Vehicle frame number identification system based on deep convolutional neural network |
| CN111950468A (en) * | 2020-08-14 | 2020-11-17 | 广东工业大学 | Traffic condition detection method and system based on convolutional neural network and storage medium |
| CN112017165A (en) * | 2020-08-19 | 2020-12-01 | 上海美沃精密仪器股份有限公司 | Lacrimal river height detection method based on deep learning |
| CN112164057A (en) * | 2020-10-09 | 2021-01-01 | 珠海格力电器股份有限公司 | Qualified label detection method, storage medium and electronic equipment |
| CN112419299A (en) * | 2020-12-04 | 2021-02-26 | 中冶建筑研究总院（深圳）有限公司 | Bolt loss detection method, device, equipment and storage medium |
| CN112419299B (en) * | 2020-12-04 | 2024-01-19 | 中冶建筑研究总院（深圳）有限公司 | Bolt missing detection method, device, equipment and storage medium |
| CN112927197A (en) * | 2021-02-19 | 2021-06-08 | 中冶建筑研究总院（深圳）有限公司 | Air conditioner external unit support corrosion detection method, device, equipment and storage medium |
| CN113012189A (en) * | 2021-03-31 | 2021-06-22 | 影石创新科技股份有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN113609900B (en) * | 2021-06-25 | 2023-09-12 | 南京信息工程大学 | Face positioning method and device for local generation, computer equipment and storage medium |
| CN113609900A (en) * | 2021-06-25 | 2021-11-05 | 南京信息工程大学 | Local generation face positioning method and device, computer equipment and storage medium |
| CN117765533A (en) * | 2024-02-22 | 2024-03-26 | 天津医科大学第二医院 | image processing method and system for oral mucosa cancer prediction |
| CN117765533B (en) * | 2024-02-22 | 2024-04-26 | 天津医科大学第二医院 | Image processing method and system for oral mucosa cancer prediction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110569769A (en) | image recognition method and device, computer equipment and storage medium | |
| CN110414507B (en) | License plate recognition method and device, computer equipment and storage medium | |
| US10817741B2 (en) | Word segmentation system, method and device | |
| WO2021027336A1 (en) | Authentication method and apparatus based on seal and signature, and computer device | |
| CN110569341B (en) | Method and device for configuring chat robot, computer equipment and storage medium | |
| CN108268867B (en) | License plate positioning method and device | |
| CN109766778A (en) | Invoice information input method, device, equipment and storage medium based on OCR technique | |
| CN109492642B (en) | License plate recognition method, license plate recognition device, computer equipment and storage medium | |
| CN108830275B (en) | Method and device for identifying dot matrix characters and dot matrix numbers | |
| CN111274957A (en) | Webpage verification code identification method, device, terminal and computer storage medium | |
| CN110781877B (en) | Image recognition method, device and storage medium | |
| CN111461170A (en) | Vehicle image detection method and device, computer equipment and storage medium | |
| CN111680690B (en) | Character recognition method and device | |
| CN110969154A (en) | Text recognition method and device, computer equipment and storage medium | |
| CN110942067A (en) | Text recognition method and device, computer equipment and storage medium | |
| CN111259894B (en) | Certificate information identification method and device and computer equipment | |
| CN111683202B (en) | Bill stamping method, device, equipment and storage medium | |
| CN112801923A (en) | Word processing method, system, readable storage medium and computer equipment | |
| CN111881880A (en) | Bill text recognition method based on novel network | |
| CN110956133A (en) | Training method of single character text normalization model, text recognition method and device | |
| CN114429636B (en) | Image scanning identification method and device and electronic equipment | |
| CN111008635A (en) | OCR-based multi-bill automatic identification method and system | |
| CN115170414A (en) | Knowledge distillation-based single image rain removing method and system | |
| CN111950550A (en) | Vehicle frame number identification system based on deep convolutional neural network | |
| CN112926610A (en) | Construction method of license plate image screening model and license plate image screening method |
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 |