CN113743445B - Target object identification method, device, computer equipment and storage medium - Google Patents
Target object identification method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN113743445B CN113743445B CN202110799036.0A CN202110799036A CN113743445B CN 113743445 B CN113743445 B CN 113743445B CN 202110799036 A CN202110799036 A CN 202110799036A CN 113743445 B CN113743445 B CN 113743445B
- Authority
- CN
- China
- Prior art keywords
- data
- processed
- target object
- probability
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000003860 storage Methods 0.000 title claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 102
- 238000012821 model calculation Methods 0.000 claims abstract description 36
- 238000004590 computer program Methods 0.000 claims description 20
- 238000007781 pre-processing Methods 0.000 claims description 13
- 238000005520 cutting process Methods 0.000 claims description 9
- 230000007547 defect Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 235000012431 wafers Nutrition 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The application relates to a target object identification method, a target object identification device, computer equipment and a storage medium. The method comprises the following steps: acquiring data to be processed; inputting the data to be processed into a recognition model obtained by training in advance to obtain a model calculation result; if the model calculation result represents that the data to be processed possibly comprises a target object, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value. By adopting the method, the accuracy of model prediction can be improved, so that the model can still have accurate recognition rate when recognizing and classifying flaws with unobvious characteristics or relatively small sizes.
Description
Technical Field
The present application relates to the field of artificial intelligence, and in particular, to a target object recognition method, apparatus, computer device, and storage medium.
Background
The convolutional neural network (Convolutional Neural Network, CNN) is inspired by biological natural vision recognition, has excellent performance for large-scale image processing, and has wide application in the fields of image classification, target identification, target positioning, semantic segmentation and the like.
In practical production applications, the convolutional neural network needs to use different network models to achieve the best prediction result according to the product type, the production condition, the defect type and the like. In addition, the neural network model needs to be trained by certain data to be deployed in a production environment for use, so the task of classifying images is often divided into two categories: model training and model prediction. The model training task needs to input modeling data into a neural network, and then back propagation is carried out to adjust the parameters of the neural network according to the difference between the output and the actual value, and the AI model for the production line can be generated through repeated adjustment; the model prediction task completes the quality detection flow in the production line in an auxiliary way, and the effect of the model prediction task is required to depend on the training effect of the preamble model.
However, at present, model training and model prediction tasks are all performed by using a GPU, and the GPU is dedicated to a special card, the semiconductor layer has tens to hundreds channels, and switching use needs to consume excessive manpower to perform field adjustment. For example, when defects with insignificant features or relatively small dimensions need to be identified, a model change is required, which requires a lot of manpower.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a target object recognition method, apparatus, computer device, and storage medium that can still have a precise recognition rate when recognizing and classifying flaws with insignificant features or relatively small dimensions.
A method of target object identification, the method comprising:
acquiring data to be processed;
Inputting the data to be processed into a recognition model obtained by training in advance to obtain a model calculation result;
If the model calculation result represents that the data to be processed possibly comprises a target object, acquiring a first probability of classification corresponding to the target object in the data to be processed;
and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the data to be processed is input into the recognition model obtained by pre-training to obtain a model calculation result, the method further includes:
If the model calculation result represents that the data to be processed does not comprise the target object, acquiring a second probability that the data to be processed belongs to other preset classifications;
When the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed;
and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the obtaining the second probability that the data to be processed belongs to the remaining preset categories, the method further includes:
and outputting the result that the data to be processed belongs to the rest preset classification when the second probability is larger than or equal to a second preset value.
In one embodiment, before the data to be processed is input into the recognition model obtained by pre-training to obtain the model calculation result, the method further includes:
acquiring training data;
Preprocessing the training data to obtain newly added data;
training the training data and the newly added data to obtain an identification model.
In one embodiment, the preprocessing the training data to obtain new data includes:
and cutting and/or amplifying the training data according to the category of the training data to obtain newly added data.
A target object recognition device, the device comprising:
The data acquisition module to be processed is used for acquiring the data to be processed;
the model training module is used for inputting the data to be processed into a recognition model obtained through pre-training so as to obtain a model calculation result;
The first probability obtaining module is used for obtaining a first probability of classification corresponding to the target object in the data to be processed if the model calculation result represents that the data to be processed possibly comprises the target object;
and the output module is used for outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, the apparatus further comprises:
The second probability obtaining module is used for obtaining the second probability that the data to be processed belongs to the rest preset classifications if the model calculation result represents that the data to be processed does not comprise the target object;
the first probability obtaining module is further configured to obtain a first probability of classification corresponding to the target object in the data to be processed when the second probability is smaller than a second preset value;
The output module is further configured to output a classification of the target object included in the data to be processed when the first probability is greater than a first preset value.
In one embodiment, the output module is further configured to output a result that the data to be processed belongs to the remaining preset classifications when the second probability is greater than or equal to a second preset value.
A computer device comprising a memory storing a computer program and a processor implementing the steps of the method of any of the embodiments described above when the computer program is executed.
A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method of any of the embodiments described above.
According to the target object identification method, the target object identification device, the computer equipment and the storage medium, the original category and the target object category are distinguished through the identification model, and after the fact that the target object possibly is included in the data to be processed is judged, judgment is carried out according to the first probability of the category of the target object in the data to be processed, so that accuracy of a result is guaranteed.
Drawings
FIG. 1 is a diagram of an application environment for a target object recognition method in one embodiment;
FIG. 2 is a flow chart of a method for identifying a target object in one embodiment;
FIG. 3 is a flowchart of a method for identifying a target object according to another embodiment;
FIG. 4 is a flow chart of a model training step in one embodiment;
FIG. 5 is a block diagram of a target object recognition device in one embodiment;
fig. 6 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The target object identification method provided by the application can be applied to an application environment shown in figure 1. The terminal 102 communicates with each machine 104 through a network, wherein the terminal 102 may acquire to-be-processed data uploaded by each machine 104 or the network, and input the to-be-processed data into an identification model obtained by training in advance to obtain a model calculation result, where if the model calculation result represents that the to-be-processed data may include a target object, a first probability that the to-be-processed data includes the target object is acquired; and outputting a result of the target object included in the data to be processed when the first probability is larger than a first preset value.
The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, among others.
In one embodiment, as shown in fig. 2, a target object recognition method is provided, and the method is applied to the terminal in fig. 1 for illustration, and includes the following steps:
S202: and obtaining data to be processed.
Specifically, the data to be processed may refer to the machine or the data uploaded by the web page, which may be displayed in the form of a picture, that is, a picture of a wafer on the machine acquired by the image acquisition device on the machine, or a picture of a wafer uploaded by the web page.
S204: and inputting the data to be processed into a recognition model obtained by pre-training to obtain a model calculation result.
Specifically, in the defect pictures with the same proportion size, defects with unobvious features or relatively small sizes are classified, wherein in order to ensure the accuracy of the identification model, training data of the identification model are improved, original training data are cut and amplified according to various types of labels and then used as new training data by a terminal, the new training data are defined to correspond to the new types, the training data comprise the original training data and the new training data, and the original training data and the new training data are trained to obtain the identification model, so that the identification model can identify the original types of defects with the original sizes and can classify the defects with unobvious features or relatively small sizes. Whether flaws with unobvious characteristics or relatively small sizes possibly exist in the data to be processed, namely whether target objects possibly exist or not is judged by introducing new categories, if so, the flaws with unobvious characteristics or relatively small sizes are continuously classified according to the original flaw categories, so that the accuracy of flaw classification is ensured.
The terminal inputs the data to be processed into the recognition model, so that the recognition model can process the data to be processed to obtain the probability of each classification of the data to be processed, namely a model calculation result.
S206: and if the model calculation result represents that the data to be processed possibly comprises the target object, acquiring a first probability of classification corresponding to the target object in the data to be processed.
Specifically, the target object is a flaw with insignificant characteristics or relatively small size, that is, corresponds to a newly introduced category in the training data.
The terminal inputs the data to be processed into the recognition model, so that the probability that the data to be processed belongs to each category can be obtained, and if the data to be processed is input into the newly introduced category, namely the data to be processed possibly comprises flaws with unobvious characteristics or relatively small sizes, the terminal obtains the first probability of the corresponding classification of the target object in the data to be processed.
S208: and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
Specifically, the first preset value may be preset, which is used for representing the confidence of the category to which the flaw with the unobvious existing characteristic or the relatively smaller size belongs, and if the first probability is greater than the first preset value, outputting the classification of the target object included in the data to be processed. Further, when the terminal determines that the data to be processed possibly includes the target object, the terminal takes as output one category with the highest probability of each category in the target category set.
The first preset value may be obtained according to an updated training sample, for example, the updated training sample is clustered to obtain a probability of the updated training sample belonging to the target classification, that is, including a flaw with an insignificant feature or a relatively small size, and the probability is taken as the first preset value, and in other embodiments, the first preset value may also be obtained by other manners, which is not limited specifically herein.
According to the target object identification method, the original category and the target object category are distinguished through the identification model, and after the target object is possibly included in the data to be processed is judged, judgment is further carried out according to the first probability that the target object is included in the data to be processed, so that accuracy of a result is guaranteed.
In one embodiment, after the data to be processed is input into the recognition model obtained by pre-training to obtain a model calculation result, the method further includes: if the model calculation result represents that the data to be processed does not comprise the target object, acquiring a second probability that the data to be processed belongs to other preset classifications; when the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the obtaining the second probability that the data to be processed belongs to the remaining preset categories, the method further includes: and outputting the result that the data to be processed belongs to the rest preset classification when the second probability is larger than or equal to a second preset value.
Specifically, in this embodiment, when the category to which the maximum probability belongs in the model calculation result is not a flaw with unobvious features or relatively small size, but is an original category, the terminal obtains a second probability that the data to be processed belongs to the original category, and determines that the data to be processed belongs to the original category when the second probability is greater than or equal to a second preset value.
The terminal acquires a second probability that the data to be processed belongs to the original category, and determines that the data to be processed does not belong to the original category when the second probability is smaller than a second preset value, and the data to be processed may include flaws with insignificant characteristics or relatively small sizes so as to perform secondary judgment. I.e., continue to determine the original classification corresponding to the flaw with the insignificant feature or relatively small size.
The second preset value is also preset by the user, and the setting manner of the second preset value can be similar to that of the first preset value, and is not repeated herein, the first preset value is a threshold value for classifying flaws with larger sizes, and the second preset value is a threshold value for classifying flaws with insignificant features or relatively smaller sizes, wherein the classification can refer to the original classification, and the classification of flaws with insignificant features or relatively smaller sizes is introduced only in the first classification, so that when the first classification determines that the to-be-processed data belongs to the flaw classification with insignificant features or relatively smaller sizes, the flaws with insignificant features or relatively smaller sizes are further classified.
In order to fully understand the target object recognition method of the present application, referring to fig. 3, fig. 3 is a schematic diagram of another embodiment of the target object recognition method, wherein in this embodiment, the object is to enable a detection model of flaw classification to have accurate recognition rate when performing recognition classification on flaws with insignificant features or relatively small sizes in flaw pictures with the same scale. For this purpose, new classes, namely flaw classes with insignificant features or relatively small dimensions, hereinafter referred to as target classes, are introduced.
In practical application, firstly, data to be processed is obtained, then the data to be processed is input into an identification model to obtain a model calculation result, and the model calculation result is used for judging the classification to which the data to be processed belongs and the probability of each classification, wherein the maximum probability is recorded as an initial classification result.
And the terminal judges whether the initial classification result belongs to target classification, namely whether flaws with unobvious features or relatively small sizes exist in the data to be processed, if the initial classification result does not belong to the target classification, judging whether the probability of the corresponding classification is larger than a second preset value, namely a default preset value of the original class, if the probability is larger than the second preset value, the default preset value of the original class, indicating that the data to be processed belongs to the original class, otherwise, judging that the data to be processed possibly has the flaws with unobvious features or relatively small sizes, and continuing to judge the next step.
If the initial classification result indicates that the data to be processed belongs to target classification, namely that flaws with unobvious characteristics or relatively small sizes exist in the data to be processed, the next judgment is also carried out.
The next step of judging is that the terminal obtains the probability that the data to be processed belongs to the target classification, namely the probability that each type of flaw with unobvious characteristics or relatively small size exists in the data to be processed, and judges whether the probability is larger than a second preset value of the corresponding probability, namely the probability is larger than a preset value of the classification aiming at the flaw with unobvious characteristics or relatively small size. If yes, outputting the category of flaws with unobvious characteristics or relatively smaller sizes in the data to be processed, otherwise, outputting the data to be processed, or directly judging that the flaws exist in the data to be processed.
The target object identification method in the embodiment improves the sensitivity to small flaws and improves the accuracy of model identification and classification. Can effectively identify and classify the defect features difficult to be found by manpower, and optimize the overall semiconductor defect classification system.
In one embodiment, before the data to be processed is input into the recognition model obtained by pre-training to obtain the model calculation result, the method further includes: acquiring training data; preprocessing the training data to obtain newly added data; training the training data and the newly added data to obtain an identification model.
In one embodiment, the preprocessing the training data to obtain new data includes: and cutting and/or amplifying the training data according to the category of the training data to obtain newly added data.
Specifically, referring to fig. 4, the training data is training data of an original recognition model, and in order to improve recognition accuracy of the recognition model, the original training data is processed to obtain new data, so that model training is performed according to the training data and the new data to obtain the recognition model.
Specifically, the processing of the original training data may be cutting and/or amplifying, for example, the terminal cuts the defective area according to the type of the original training data, and amplifies the defective area as a new type, where the amplified picture is consistent with the original picture in size, so as to ensure that the sizes of the pictures input into the recognition model are the same. The new class is defined as a small flaw or flaw-free class, i.e. the flaw needs to be amplified, and is defined as a target class set. In the prediction, the target class set can be used for judging whether the classification result of the primary prediction accords with the small flaw or flaw-free class. If the model classification meets the target class, the prediction result classifies the picture through a more strict threshold value, and the accuracy of model classification can be effectively improved through the mode.
Wherein the training process of the model comprises: each category of the input training data set is additionally added with a new category by cutting or amplifying and the like. The additional new categories are combined with the original training data into a new data set. Model training for deep learning is performed on the merged new data set.
Wherein the following detailed description is presented to make the application more apparent to those skilled in the art:
Because the sizes of flaws in wafers are different, the target object identification method provided by the application can still have accurate identification rate when identification and classification are carried out on flaws with unobvious characteristics or relatively small sizes in flaw pictures with the same scale by using the identification model for flaw classification.
After each class of the original training data set is cut and amplified, the original training data set is added into a new data class set, the new data class set is defined as a target class set, and each class of the original training data set is defined as an original class set.
The identification model trained by the data of the original class set and the target class set is divided into three steps in the application of flaw classification, wherein the first step is to predict the scores of the classes in the original class set, and the highest predicted score of the classes is the initial classification result; secondly, judging whether the classification result belongs to a target class, namely a small flaw or flaw-free class; and thirdly, judging whether the prediction score of the category has enough confidence, if the classification result does not belong to the small flaw or flaw-free category and has enough confidence on the prediction score of the classification result, the initial classification result is the final output, otherwise, the classification result with the highest prediction score of each category in the target category set and enough confidence is the final output.
The target object identification method mainly comprises the following steps: a model with a set of original classes and a set of target classes can be trained. The model is applied to classify flaws and obtain an initial predictive classification and predictive score. Judging whether the prediction score has enough confidence and whether the classification result is a flawless or small-sized flaw type.
It should be understood that, although the steps in the flowcharts of fig. 2-4 are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 2-4 may include multiple steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor does the order in which the steps or stages are performed necessarily performed in sequence, but may be performed alternately or alternately with at least a portion of the steps or stages in other steps or other steps.
In one embodiment, as shown in fig. 5, there is provided a target object recognition apparatus including: the system comprises a data acquisition module 100 to be processed, a model training module 200, a first probability acquisition module 300 and an output module 400, wherein:
a data to be processed acquisition module 100 for acquiring data to be processed;
The model training module 200 is configured to input the data to be processed into an identification model obtained by training in advance, so as to obtain a model calculation result;
the first probability obtaining module 300 is configured to obtain a first probability of a classification corresponding to a target object in the data to be processed if the model calculation result represents that the data to be processed may include the target object;
And the output module 400 is configured to output the classification of the target object included in the data to be processed when the first probability is greater than a first preset value.
In one embodiment, the apparatus further comprises:
The second probability obtaining module is used for obtaining the second probability that the data to be processed belongs to the rest preset classifications if the model calculation result represents that the data to be processed does not comprise the target object;
the first probability obtaining module 300 is further configured to obtain a first probability of a classification corresponding to the target object in the data to be processed when the second probability is smaller than a second preset value;
The output module 400 is further configured to output a classification of the target object included in the data to be processed when the first probability is greater than a first preset value.
In one embodiment, the output module 400 is further configured to output a result that the data to be processed belongs to the remaining preset classifications when the second probability is greater than or equal to a second preset value.
In one embodiment, the target object recognition apparatus further includes:
The training data acquisition module is used for acquiring training data;
the preprocessing module is used for preprocessing the training data to obtain newly added data;
and the training module is used for training the training data and the newly added data to obtain an identification model.
In one embodiment, the preprocessing module is configured to perform cutting and/or amplifying processing on the training data according to the class of the training data to obtain the newly added data.
For specific limitations of the target object recognition apparatus, reference may be made to the above limitations of the target object recognition method, and no further description is given here. The respective modules in the above-described target object recognition apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a terminal, and the internal structure of which may be as shown in fig. 6. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device 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 includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a target object recognition method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the 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 stored therein a computer program, the processor when executing the computer program performing the steps of: acquiring data to be processed; inputting the data to be processed into a recognition model obtained by training in advance to obtain a model calculation result; if the model calculation result represents that the data to be processed possibly comprises a target object, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the data to be processed is input into the recognition model obtained by pre-training and implemented when the processor executes the computer program, the method further includes: if the model calculation result represents that the data to be processed does not comprise the target object, acquiring a second probability that the data to be processed belongs to other preset classifications; when the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the obtaining the second probability that the data to be processed belongs to the remaining preset categories, the method further includes: and outputting the result that the data to be processed belongs to the rest preset classification when the second probability is larger than or equal to a second preset value.
In one embodiment, before the data to be processed is input into the recognition model obtained by pre-training and implemented when the processor executes the computer program, the method further comprises: acquiring training data; preprocessing the training data to obtain newly added data; training the training data and the newly added data to obtain an identification model.
In one embodiment, the preprocessing of the training data to obtain new data implemented when the processor executes the computer program comprises: and cutting and/or amplifying the training data according to the category of the training data to obtain newly added data.
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 data to be processed; inputting the data to be processed into a recognition model obtained by training in advance to obtain a model calculation result; if the model calculation result represents that the data to be processed possibly comprises a target object, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the inputting the data to be processed into the recognition model obtained by pre-training, when the computer program is executed by the processor, the method further includes: if the model calculation result represents that the data to be processed does not comprise the target object, acquiring a second probability that the data to be processed belongs to other preset classifications; when the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed; and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
In one embodiment, after the obtaining the second probability that the data to be processed belongs to the remaining preset categories, the method further includes: and outputting the result that the data to be processed belongs to the rest preset classification when the second probability is larger than or equal to a second preset value.
In one embodiment, before the inputting the data to be processed into the recognition model obtained by pre-training, the computer program when executed by the processor further comprises: acquiring training data; preprocessing the training data to obtain newly added data; training the training data and the newly added data to obtain an identification model.
In one embodiment, the preprocessing of the training data to obtain new data, which is implemented when the computer program is executed by the processor, comprises: and cutting and/or amplifying the training data according to the category of the training data to obtain newly added data.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, or the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration, and not limitation, RAM can be in various forms such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples illustrate only a few embodiments of the application, which are described in detail and are not to be construed as limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (4)
1. A method of identifying a target object, the method comprising:
acquiring training data; preprocessing the training data to obtain newly added data; training the training data and the newly added data to obtain an identification model; cutting and/or amplifying the training data according to the category of the training data to obtain newly added data;
Acquiring data to be processed; inputting the data to be processed into a recognition model obtained by training in advance to obtain a model calculation result;
If the model calculation result represents that the data to be processed possibly comprises a target object, acquiring a first probability of classification corresponding to the target object in the data to be processed;
Outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value;
if the model calculation result represents that the data to be processed does not comprise the target object, acquiring a second probability that the data to be processed belongs to other preset classifications; outputting a result that the data to be processed belongs to the rest preset classifications when the second probability is larger than or equal to a second preset value; when the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed;
and outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value.
2. A target object recognition apparatus, the apparatus comprising:
The data acquisition module to be processed is used for acquiring the data to be processed;
The recognition model module is used for acquiring training data; preprocessing the training data to obtain newly added data; training the training data and the newly added data to obtain an identification model; cutting and/or amplifying the training data according to the category of the training data to obtain newly added data;
the model training module is used for inputting the data to be processed into a recognition model obtained through pre-training so as to obtain a model calculation result;
The first probability obtaining module is used for obtaining a first probability of classification corresponding to the target object in the data to be processed if the model calculation result represents that the data to be processed possibly comprises the target object; when the second probability is smaller than a second preset value, acquiring a first probability of classification corresponding to the target object in the data to be processed;
The second probability obtaining module is used for obtaining the second probability that the data to be processed belongs to the rest preset classifications if the model calculation result represents that the data to be processed does not comprise the target object;
The output module is used for outputting the classification of the target object included in the data to be processed when the first probability is larger than a first preset value; and outputting the result that the data to be processed belongs to the rest preset classification when the second probability is larger than or equal to a second preset value.
3. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of claim 1 when executing the computer program.
4. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110799036.0A CN113743445B (en) | 2021-07-15 | 2021-07-15 | Target object identification method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110799036.0A CN113743445B (en) | 2021-07-15 | 2021-07-15 | Target object identification method, device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113743445A CN113743445A (en) | 2021-12-03 |
| CN113743445B true CN113743445B (en) | 2024-06-04 |
Family
ID=78728673
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110799036.0A Active CN113743445B (en) | 2021-07-15 | 2021-07-15 | Target object identification method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113743445B (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105654121A (en) * | 2016-04-11 | 2016-06-08 | 李云栋 | Complex jacquard fabric defect detection method based on deep learning |
| CN109509180A (en) * | 2018-10-24 | 2019-03-22 | 东华大学 | Metal button flaw detection method based on machine vision |
| CN109741296A (en) * | 2018-11-28 | 2019-05-10 | 歌尔股份有限公司 | Product quality detection method and device |
| CN110779928A (en) * | 2019-11-19 | 2020-02-11 | 汪科道 | Defect detection device and method |
| CN110796034A (en) * | 2019-10-12 | 2020-02-14 | 北京达佳互联信息技术有限公司 | Target object identification method, device, equipment and medium |
| CN110827157A (en) * | 2019-11-05 | 2020-02-21 | 泰康保险集团股份有限公司 | Data processing method and device, storage medium and electronic equipment |
| CN111192262A (en) * | 2020-01-03 | 2020-05-22 | 腾讯云计算（北京）有限责任公司 | Product defect classification method, device, equipment and medium based on artificial intelligence |
| CN111210024A (en) * | 2020-01-14 | 2020-05-29 | 深圳供电局有限公司 | Model training method and device, computer equipment and storage medium |
| CN111222423A (en) * | 2019-12-26 | 2020-06-02 | 深圳供电局有限公司 | Target identification method and device based on operation area and computer equipment |
| CN111507420A (en) * | 2020-04-22 | 2020-08-07 | 上海眼控科技股份有限公司 | Tire information acquisition method, tire information acquisition device, computer device, and storage medium |
| CN111595850A (en) * | 2020-04-27 | 2020-08-28 | 平安科技（深圳）有限公司 | Slice defect detection method, electronic device and readable storage medium |
| DE102019110721A1 (en) * | 2019-04-25 | 2020-10-29 | Carl Zeiss Industrielle Messtechnik Gmbh | WORKFLOW FOR TRAINING A CLASSIFICATOR FOR QUALITY INSPECTION IN MEASUREMENT TECHNOLOGY |
| WO2021068513A1 (en) * | 2019-10-12 | 2021-04-15 | 平安科技（深圳）有限公司 | Abnormal object recognition method and apparatus, medium, and electronic device |
| CN112819815A (en) * | 2021-02-25 | 2021-05-18 | 汇鸿智能科技(辽宁)有限公司 | Metallographic structure identification method and device, computer equipment and storage medium |
| US11035803B1 (en) * | 2020-02-26 | 2021-06-15 | Applied Materials Israel Ltd. | Multi-perspective examination of a specimen |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7215811B2 (en) * | 2000-11-22 | 2007-05-08 | Osama Moselhi | Method and apparatus for the automated detection and classification of defects in sewer pipes |
| US11037286B2 (en) * | 2017-09-28 | 2021-06-15 | Applied Materials Israel Ltd. | Method of classifying defects in a semiconductor specimen and system thereof |
-
2021
- 2021-07-15 CN CN202110799036.0A patent/CN113743445B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105654121A (en) * | 2016-04-11 | 2016-06-08 | 李云栋 | Complex jacquard fabric defect detection method based on deep learning |
| CN109509180A (en) * | 2018-10-24 | 2019-03-22 | 东华大学 | Metal button flaw detection method based on machine vision |
| CN109741296A (en) * | 2018-11-28 | 2019-05-10 | 歌尔股份有限公司 | Product quality detection method and device |
| DE102019110721A1 (en) * | 2019-04-25 | 2020-10-29 | Carl Zeiss Industrielle Messtechnik Gmbh | WORKFLOW FOR TRAINING A CLASSIFICATOR FOR QUALITY INSPECTION IN MEASUREMENT TECHNOLOGY |
| CN110796034A (en) * | 2019-10-12 | 2020-02-14 | 北京达佳互联信息技术有限公司 | Target object identification method, device, equipment and medium |
| WO2021068513A1 (en) * | 2019-10-12 | 2021-04-15 | 平安科技（深圳）有限公司 | Abnormal object recognition method and apparatus, medium, and electronic device |
| CN110827157A (en) * | 2019-11-05 | 2020-02-21 | 泰康保险集团股份有限公司 | Data processing method and device, storage medium and electronic equipment |
| CN110779928A (en) * | 2019-11-19 | 2020-02-11 | 汪科道 | Defect detection device and method |
| CN111222423A (en) * | 2019-12-26 | 2020-06-02 | 深圳供电局有限公司 | Target identification method and device based on operation area and computer equipment |
| CN111192262A (en) * | 2020-01-03 | 2020-05-22 | 腾讯云计算（北京）有限责任公司 | Product defect classification method, device, equipment and medium based on artificial intelligence |
| CN111210024A (en) * | 2020-01-14 | 2020-05-29 | 深圳供电局有限公司 | Model training method and device, computer equipment and storage medium |
| US11035803B1 (en) * | 2020-02-26 | 2021-06-15 | Applied Materials Israel Ltd. | Multi-perspective examination of a specimen |
| CN111507420A (en) * | 2020-04-22 | 2020-08-07 | 上海眼控科技股份有限公司 | Tire information acquisition method, tire information acquisition device, computer device, and storage medium |
| CN111595850A (en) * | 2020-04-27 | 2020-08-28 | 平安科技（深圳）有限公司 | Slice defect detection method, electronic device and readable storage medium |
| CN112819815A (en) * | 2021-02-25 | 2021-05-18 | 汇鸿智能科技(辽宁)有限公司 | Metallographic structure identification method and device, computer equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| "核安全壳表面裂缝自动检测关键技术研究";黄晶晶;《中国优秀硕士学位论文全文数据库 信息科技辑》;20200615;全文 * |
| "Wafer Map Defect Pattern Classification and Image Retrieval Using Convolutional Neural Network";Takeshi Nakazawa等;《IEEE》;20180118;全文 * |
| 高分二号遥感影像提取冬小麦空间分布;宋德娟;张承明;杨晓霞;李峰;韩颖娟;高帅;董海燕;;遥感学报(第05期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113743445A (en) | 2021-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109285105B (en) | Watermark detection method, watermark detection device, computer equipment and storage medium | |
| CN112926654B (en) | Pre-labeling model training and certificate pre-labeling method, device, equipment and medium | |
| CN109308488B (en) | Mammary gland ultrasonic image processing device, method, computer equipment and storage medium | |
| CN109697468B (en) | Sample image labeling method and device and storage medium | |
| CN111368636B (en) | Object classification method, device, computer equipment and storage medium | |
| CN111274926B (en) | Image data screening method, device, computer equipment and storage medium | |
| CN110889437B (en) | Image processing method and device, electronic equipment and storage medium | |
| CN110705489B (en) | Training method and device for target recognition network, computer equipment and storage medium | |
| CN111931809A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN111259823A (en) | Pornographic image identification method based on convolutional neural network | |
| CN109754077B (en) | Network model compression method and device of deep neural network and computer equipment | |
| CN113435182A (en) | Method, device and equipment for detecting conflict of classification labels in natural language processing | |
| JP2022185143A (en) | Text detection method, and text recognition method and device | |
| CN111768405A (en) | Method, device, equipment and storage medium for processing annotated image | |
| CN115171125A (en) | Data anomaly detection method | |
| CN113326887B (en) | Text detection method, device and computer equipment | |
| CN116415020A (en) | Image retrieval method, device, electronic equipment and storage medium | |
| CN114511857A (en) | OCR recognition result processing method, device, equipment and storage medium | |
| CN113780116A (en) | Invoice classification method and device, computer equipment and storage medium | |
| CN117422934A (en) | Abnormal cell detection method, abnormal cell detection system, abnormal cell detection computer device, and storage medium | |
| CA3179311A1 (en) | Identifying claim complexity by integrating supervised and unsupervised learning | |
| CN113283396A (en) | Target object class detection method and device, computer equipment and storage medium | |
| CN113743445B (en) | Target object identification method, device, computer equipment and storage medium | |
| CN113743448B (en) | Model training data acquisition method, model training method and device | |
| CN114330542A (en) | Sample mining method and device based on target detection and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |