CN108491866B - Pornographic picture identification method, electronic device and readable storage medium - Google Patents
Pornographic picture identification method, electronic device and readable storage medium Download PDFInfo
- Publication number
- CN108491866B CN108491866B CN201810184583.6A CN201810184583A CN108491866B CN 108491866 B CN108491866 B CN 108491866B CN 201810184583 A CN201810184583 A CN 201810184583A CN 108491866 B CN108491866 B CN 108491866B
- Authority
- CN
- China
- Prior art keywords
- picture
- pornographic
- identified
- pictures
- preset
- 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
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Abstract
The invention relates to a pornographic picture identification method, an electronic device and a readable storage medium, wherein the method comprises the following steps: after receiving a picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, and the convolutional neural network model is obtained by utilizing the sample image of the preset image category to train in advance; and judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value. The invention can accurately and effectively judge whether the picture to be identified belongs to the pornographic picture or not; and need not artifical the detection, can carry out pornographic picture appraisal voluntarily, effectively improve detection efficiency.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a pornographic picture identification method, an electronic device and a readable storage medium.
Background
At present, for large-scale internet financial enterprises, a large number of business pictures are involved, and pornographic pictures possibly are mixed in the business pictures, so that the influence is severe, and effective identification and elimination are required. The traditional erotic picture identification mode is that the pornographic pictures are screened out by manually checking the service pictures one by one, and the manual detection has high cost, time consumption and lower efficiency.
Disclosure of Invention
The invention aims to provide a pornographic picture identification method, an electronic device and a readable storage medium, and aims to improve the efficiency of identifying pornographic pictures.
In order to achieve the above object, the present invention provides a pornographic picture identifying method, comprising:
after receiving a picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, and the convolutional neural network model is obtained by utilizing the sample image of the preset image category to train in advance;
and judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value.
Preferably, the picture formats include a black-and-white format and a non-black-and-white format, and the preset picture categories include pornographic pictures and non-pornographic pictures;
the step of outputting the probability value that the picture to be identified belongs to one or more preset picture categories specifically comprises: outputting the probability value of the pornographic picture belonging to the picture to be identified and the probability value of the non-pornographic picture;
the step of judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format and the probability value of the picture to be identified specifically comprises the following steps:
reducing the probability value that the picture with the black-and-white format belongs to the pornographic picture;
and judging whether the black-and-white picture belongs to the pornographic picture or not according to the reduced probability value, and judging whether the non-black-and-white picture belongs to the pornographic picture or not according to the probability value.
Preferably, the step of determining whether the black-and-white picture belongs to the pornographic picture according to the reduced probability value specifically includes:
comparing the reduced probability value with a first preset threshold value, and if the reduced probability value is greater than the preset first threshold value, judging that the picture to be identified belongs to a pornographic picture;
and if the lowered probability value is less than or equal to the preset first threshold value, judging that the picture to be identified does not belong to the pornographic picture.
Preferably, the picture formats include a black-and-white format and a non-black-and-white format, and the preset picture categories include pornographic pictures, sexy pictures and normal pictures;
the step of outputting the probability value that the picture to be identified belongs to one or more preset picture categories specifically comprises the following steps: outputting the probability value of the pornographic picture belonging to the picture to be identified, the probability value of the sexy picture and the probability value of the normal picture;
the step of judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value specifically comprises the following steps:
reducing the probability value that the picture with the black-and-white format belongs to the pornographic picture;
and judging whether the black-and-white picture belongs to the pornographic picture or not according to the reduced probability value, and judging whether the non-black-and-white picture belongs to the pornographic picture or not according to the probability value.
Preferably, the step of determining whether the black-and-white picture belongs to the pornographic picture according to the reduced probability value specifically includes:
comparing the reduced probability value with a second preset threshold value, and if the reduced probability value is greater than the preset second threshold value, judging that the picture to be identified belongs to a pornographic picture;
and if the lowered probability value is less than or equal to the preset second threshold value, judging that the picture to be identified does not belong to the pornographic picture.
Preferably, the training step of the image classification model is as follows:
A. collecting a preset number of pornographic pictures, sexy pictures and normal pictures as sample pictures, and marking corresponding picture categories on each sample picture;
B. preprocessing each sample picture;
C. dividing the preprocessed sample picture into a training set with a first proportion and a verification set with a second proportion;
D. training a convolutional neural network model by using the training set;
E. verifying the accuracy of the trained convolutional neural network model by using the verification set, finishing the training if the accuracy is greater than or equal to a preset accuracy, and taking the trained convolutional neural network model as a picture classification model; if the accuracy is less than the predetermined accuracy, the sample picture is collected again and the above step B, C, D, E is repeated.
Preferably, when the picture to be authenticated is a dynamic picture, the method further comprises the following steps:
cutting the picture to be identified into a plurality of pictures, detecting the picture format of each cut picture, respectively identifying the plurality of cut pictures by using a pre-trained picture classification model, and outputting the probability value of each cut picture belonging to each preset picture category;
and judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format and the probability value of each picture of the cut frame.
Preferably, the step of determining whether the picture to be identified belongs to the pornographic picture according to the picture format and the probability value of each picture of the cut frame specifically includes:
judging whether pictures belonging to pornographic pictures exist in the cut-frame pictures or not according to the picture format and the probability value of each picture of the cut-frame;
if the picture to be identified belongs to the pornographic picture, judging that the picture to be identified belongs to the pornographic picture; otherwise, judging that the picture to be identified does not belong to the pornographic picture; or
Judging whether more than two pictures with pornographic picture probability values larger than a preset value in the pictures of the cut frames exist according to the picture formats of the pictures of the cut frames and the probability values;
if so, judging that the picture to be identified belongs to the pornographic picture; otherwise, judging that the picture to be identified does not belong to the pornographic picture.
In addition, to achieve the above object, the present invention further provides an electronic device, which includes a memory and a processor, wherein the memory stores a pornographic picture identification system capable of running on the processor, and the pornographic picture identification system, when executed by the processor, implements the following steps:
after receiving a picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, and the convolutional neural network model is obtained by utilizing the sample image of the preset image category to train in advance;
and judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value.
Further, to achieve the above object, the present invention also provides a computer readable storage medium storing a pornographic picture evaluation system, the pornographic picture evaluation system being executable by at least one processor to cause the at least one processor to perform the steps of the pornographic picture evaluation method as described above.
According to the pornographic picture identification method, system and readable storage medium, pictures to be identified are identified through a pre-trained picture classification model, and the probability that the pictures to be identified belong to each preset picture category is output; detecting the picture format of the picture to be identified; and judging whether the picture to be identified belongs to the pornographic picture or not based on a preset rule according to the probability that the picture to be identified belongs to each preset picture category and the picture format of the picture to be identified. The invention can identify the probability that the picture to be identified belongs to one or more preset picture categories, and carry out comprehensive identification by combining the picture format of the picture to be identified, thereby more accurately and effectively judging whether the picture to be identified belongs to the pornographic picture or not. Moreover, the pornographic picture identification can be automatically carried out without manual detection, and the detection efficiency is effectively improved.
Drawings
FIG. 1 is a schematic diagram of an operating environment of a pornographic picture identification system 10 according to a preferred embodiment of the present invention;
fig. 2 is a flowchart illustrating an embodiment of a pornographic picture identifying method according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention 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 invention and are not intended to limit the invention. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
It should be noted that the description relating to "first", "second", etc. in the present invention is for descriptive purposes only and is not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In addition, technical solutions between various embodiments may be combined with each other, but must be realized by a person skilled in the art, and when the technical solutions are contradictory or cannot be realized, such a combination should not be considered to exist, and is not within the protection scope of the present invention.
The invention provides a pornographic picture identification system. Please refer to fig. 1, which is a schematic diagram illustrating an operating environment of a pornographic picture identification system 10 according to a preferred embodiment of the present invention.
In the present embodiment, the pornographic picture identification system 10 is installed and operated in the electronic device 1. The electronic device 1 may include, but is not limited to, a memory 11, a processor 12, and a display 13. Fig. 1 only shows the electronic device 1 with components 11-13, but it is to be understood that not all of the shown components are required to be implemented, and that more or fewer components may alternatively be implemented.
The memory 11 is at least one type of readable computer storage medium, and the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a hard disk or a memory of the electronic device 1. The memory 11 may also be an external storage device of the electronic apparatus 1 in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, which are provided on the electronic apparatus 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic apparatus 1. The memory 11 is used for storing application software installed in the electronic device 1 and various types of data, such as program codes of the pornographic picture identification system 10. The memory 11 may also be used to temporarily store data that has been output or is to be output.
The processor 12 may be, in some embodiments, a Central Processing Unit (CPU), microprocessor or other data Processing chip, for running program codes stored in the memory 11 or Processing data, such as executing the pornographic picture evaluation system 10.
The display 13 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch panel, or the like in some embodiments. The display 13 is used for displaying information processed in the electronic device 1 and for displaying a visual user interface, such as the probability that the identified picture to be authenticated belongs to each preset picture category, the picture format of the picture to be authenticated, the authentication result, and the like. The components 11-13 of the electronic device 1 communicate with each other via a system bus.
Pornographic picture evaluation system 10 includes at least one computer readable instruction stored in memory 11 that is executable by processor 12 to implement embodiments of the present application.
Wherein, the pornographic picture identifying system 10 when executed by the processor 12 implements the following steps:
step S1, after receiving the picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, and the convolutional neural network model is obtained by utilizing the sample image of the preset image category to train in advance.
In this embodiment, the pornographic picture identification system receives a pornographic picture identification request including a picture to be identified sent by a user, for example, the pornographic picture identification request sent by the user through a terminal such as a mobile phone, a tablet computer, or a self-service terminal device, for example, the pornographic picture identification request sent by the user from a client installed in the terminal such as the mobile phone, the tablet computer, or the self-service terminal device in advance, or the pornographic picture identification request sent by the user from a browser system in the terminal such as the mobile phone, the tablet computer, or the self-service terminal device.
After receiving a pornographic picture identification request sent by a user, the pornographic picture identification system identifies the received picture to be identified by using a pre-trained picture classification model and outputs the probability that the picture to be identified belongs to each preset picture category (pornographic picture and non-pornographic picture). The image classification model can be trained into a model capable of effectively outputting the probability that the image to be identified belongs to each preset image category (pornographic image and non-pornographic image) by continuously performing training, learning, verification, optimization and the like by identifying a large number of preset sample images marked with different preset image categories (pornographic image and non-pornographic image). For example, the picture classification model may employ a deep Convolutional Neural Network (CNN) model or the like.
Meanwhile, the picture format of the picture to be identified can be detected, wherein the picture format comprises a black-and-white picture and a color picture. For example, by an algorithm for judging RGB values, whether the picture to be authenticated is a black-and-white picture or a color picture is judged, and the black-and-white picture or the color picture is marked. Specifically, the manner of determining whether the picture to be identified is a black-and-white picture or a color picture is as follows: if the RGB values of the pixels in the photo are all 0 or 1, the photo to be identified is judged to be a black-and-white photo, and otherwise, the photo is a color photo.
And step S2, judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value. For example, since the pornographic pictures are generally color pictures and the probability that the pornographic pictures are black-and-white pictures is low, if the picture format of the picture to be identified is judged to be black-and-white pictures, the probability that the picture to be identified belongs to the pornographic picture category output by the picture classification model is reduced, and if the probability that the picture to be identified belongs to the pornographic picture category is still smaller than a preset threshold (such as 50%) after being reduced, the picture to be identified is judged not to belong to the pornographic pictures; if the probability that the picture to be identified belongs to the pornographic picture category is lower than a preset threshold (such as 50%), judging that the picture to be identified belongs to the pornographic picture. If the picture format of the picture to be identified is judged to be a color picture, directly comparing the probability value of the picture to be identified, which is output by the picture classification model and belongs to the pornographic picture category, with a preset threshold (such as 60%), and if the probability value is smaller than the preset threshold (such as 60%), judging that the picture to be identified does not belong to the pornographic picture; if the picture to be identified is larger than or equal to the preset threshold (such as 60%), the picture to be identified is judged to belong to the pornographic picture.
Compared with the prior art, the method and the device have the advantages that the picture to be identified is identified through the pre-trained picture classification model, and the probability that the picture to be identified belongs to each preset picture category such as pornographic pictures and non-pornographic pictures is output; detecting the picture format of the picture to be identified, such as a black-and-white picture or a color picture; and judging whether the picture to be identified belongs to the pornographic picture or not based on a preset rule according to the probability that the picture to be identified belongs to each preset picture category and the picture format of the picture to be identified. Because the probability that the pornographic picture is a black-and-white picture is lower, the method can identify the probability that the picture to be identified belongs to the pornographic picture and the non-pornographic picture, and can judge whether the picture to be identified belongs to the pornographic picture more accurately and effectively by carrying out comprehensive identification by combining the picture format of the picture to be identified such as the black-and-white picture or the color picture. Moreover, the pornographic picture identification can be automatically carried out without manual detection, and the detection efficiency is effectively improved.
In an optional embodiment, based on the embodiment of fig. 1, the training steps of the pre-trained image classification model are as follows:
A. setting a corresponding preset number of sample pictures for each preset picture type, and marking the corresponding preset picture type for each sample picture; the preset picture categories comprise pornographic pictures, sexy pictures and normal pictures;
B. carrying out picture preprocessing on each sample picture to obtain a training picture to be subjected to model training;
C. dividing all training pictures into a first proportion (e.g., 75%) of training set and a second proportion of validation set (e.g., 25%);
D. training a convolutional neural network model by using the training set;
E. and verifying the accuracy of the trained convolutional neural network model by using the verification set, if the accuracy is greater than or equal to the preset accuracy, ending the training, and taking the trained convolutional neural network model as a picture classification model, or if the accuracy is less than the preset accuracy, increasing the number of sample pictures corresponding to each preset picture category and executing the step B, C, D, E again until the accuracy of the trained convolutional neural network model is greater than or equal to the preset accuracy.
Specifically, in the present embodiment, when establishing the CNN image classification model, a large number of images of different categories are prepared, for example, 10 ten thousand sample images classified into pornographic, sexy, and normal categories are prepared, and the category to which each image belongs is labeled, for example, the category to which each image belongs may be labeled as follows, 0: the pornography; 1: the sense of sex; 2: and (4) normal. Wherein, the normal pictures can include natural, artistic, dead body and other pictures; the pictures of the sexual category may include pictures defined not to be pornographic pictures but between pornography and normal, such as pictorial pictures; the pictures of the pornographic categories are defined pictures belonging to pornographic, and are the key points of picture identification in the embodiment. After the sample pictures are prepared, the sample pictures can be subjected to various pre-processing, such as being cut into training pictures with uniform size (the pictures are all resize 100X100 size) or uniform pixels. And training a Convolutional Neural Network (CNN) model with a preset model structure by using the preprocessed training pictures and the labeled categories (0: pornography, 1: sexuality and 2: normal). For example, in an alternative embodiment, the training process is as follows:
(1) establishing a convolution kernel and a deviation matrix;
(2) carrying out convolution operation by utilizing the established convolution kernel, the deviation matrix and the training picture;
(3) correcting the result of the convolution operation through the activation function of relu;
(4) pooling is a posing operation, which is a method of reducing an image while preserving most of the important information of the image. The pooling in this embodiment includes, but is not limited to, Mean pooling, Max pooling, overlappinging, L2pooling, Local Contrast Normalization, Stochasticpooling, Def-pooling, and the like.
(5) Repeating the steps (1) to (4) for 3 times;
(6) and ending the model training until the gradient of the value function of the trained model is reduced.
In addition, in the online application process of the model, if the service system finds abnormal pictures such as pornographic pictures, the found abnormal pictures can be supplemented to continue training the Convolutional Neural Network (CNN) model so as to continuously improve the identification precision of the Convolutional Neural Network (CNN) model.
After the CNN image classification model is established, the CNN image classification model can be used for image classification. The picture to be identified is input into the trained CNN picture classification model, and the probability that the picture to be identified belongs to each preset picture category, namely the probability P0 that the picture to be identified belongs to the pornographic category, the probability P1 that the picture to be identified belongs to the sexual sensation category and the probability P2 that the picture to be identified belongs to the normal category, can be output through the trained CNN picture classification model.
Meanwhile, the picture format of the picture to be identified can be judged, in the embodiment, on the basis of judging that the picture to be identified is a black-and-white picture or a color picture, whether the picture to be identified is a dynamic picture such as a picture of the type of GIF (graphic interchange format) or not can be judged, and if the picture to be identified is judged to be the dynamic picture such as the type of GIF, the picture of the type of GIF is framed into a plurality of pictures for subsequent use. Specifically, there are two ways to determine whether the picture to be identified is a dynamic picture: the method comprises the steps of judging through a file suffix name of a picture to be identified, and judging through a file header of the picture to be identified in a binary format.
After the probability (P0, P1, P2) that the picture to be identified belongs to each preset picture category (pornography, sexuality and normality) is output by utilizing the CNN picture classification model, and the format (black-and-white picture or color picture or dynamic picture) of the picture to be identified is judged, the probability (P0, P1 and P2) and the format of the picture to be identified can be integrated, and whether the picture to be identified is the pornography picture or not can be finally identified by preset rules. Specifically, the method comprises the following steps:
since the pornographic pictures are usually color pictures and the probability that the pornographic pictures are black-and-white pictures is low, if the picture to be identified is determined to be a black-and-white picture, the value of the probability P0 that the picture to be identified belongs to the pornographic category output by the CNN picture classification model is adjusted downward by 10%, for example, and if the adjusted value of P0 is smaller than a first threshold, it is determined that the picture to be identified does not belong to the pornographic category, that is, P0 × 90% < 50%, and the picture to be identified is not determined to be a pornographic picture. If the adjusted-down P0 is greater than the first threshold, the picture to be identified is determined to belong to the pornographic category, namely P0 × 90% > 50%, and the picture to be identified is determined to be the pornographic picture.
Meanwhile, the probability that the picture to be identified output by the CNN picture classification model belongs to the sexuality and normal category can be adjusted up to 10%. After the adjusted probability (P0 ', P1', P2 ') that the picture to be identified belongs to each preset picture category (pornography, sexuality and normality) is obtained, the type corresponding to the MAX (P0, P1 and P2) is the identified type, the confidence rate of the type is MAX (P0, P1 and P2), namely, the picture category (pornography, sexuality or normality) corresponding to the maximum value in (P0', P1 ', P2') is the finally identified picture category to be identified. It should be noted that, on the basis that the P0 that is turned down in the previous step is greater than the first threshold, no matter what the picture category corresponding to the maximum value in (P0 ', P1 ', P2 ') is, the finally identified category of the picture to be identified is still a pornographic picture.
If the picture to be identified is judged to be a dynamic picture, the picture to be identified, such as a GIF (graphic interchange Format) moving picture, is cut into a plurality of pictures, the probability (P0, P1, P2) of belonging to each preset picture category (pornography, sexuality and normality) is output for each frame of the cut picture by using a CNN picture classification model, each frame of the picture is judged to belong to a black-white picture or a color picture, the pornography identification is carried out on each frame of the picture by using the rule method, and whether the picture to be identified, such as the GIF moving picture, belongs to a pornography picture or not is comprehensively judged according to the identification result of each frame of the picture to be identified, such as the GIF moving picture. For example, in an alternative embodiment, if one frame of picture is identified as a pornographic picture, it is directly determined that the picture to be identified, such as the GIF motion picture, belongs to the pornographic picture. In another alternative embodiment, if both the P0' of two frames of pictures are greater than the second threshold (60% or 70%), then the picture to be authenticated, such as the GIF motion picture, is determined to belong to a pornographic picture, and so on.
The final returned result in this embodiment includes: whether the picture to be identified is a pornographic picture or not and the probability (P0 ', P1 ', P2 ') that the picture to be identified belongs to each preset picture category (pornography, sexuality and normality). In different service scenes in practical application, the first threshold value and the second threshold value can be manually set according to needs so as to be suitable for strict or loose pornographic picture identification scenes. As in some critical scenes, pictures belonging to the sexually sensitive class can be identified according to the probabilities (P0 ', P1 ', P2 ').
It should be noted that, when the picture formats include a black-and-white format and a non-black-and-white format, and the preset picture categories include pornographic pictures and non-pornographic pictures, the probability value that the picture to be identified belongs to the pornographic pictures and the probability value of the non-pornographic pictures are output; reducing the probability value that the picture with the black and white format belongs to the pornographic picture; comparing the reduced probability value with a first preset threshold value, and if the reduced probability value is greater than the preset first threshold value, judging that the picture to be identified belongs to a pornographic picture; and if the lowered probability value is less than or equal to the preset first threshold value, judging that the picture to be identified does not belong to the pornographic picture.
When the picture formats comprise a black-and-white format and a non-black-and-white format, and the preset picture categories comprise pornographic pictures, sexually sensed pictures and normal pictures, outputting the probability value of the pornographic pictures, the probability value of the sexually sensed pictures and the probability value of the normal pictures of the picture to be identified; reducing the probability value that the picture with the black and white format belongs to the pornographic picture; comparing the reduced probability value with a second preset threshold value, and if the reduced probability value is greater than the preset second threshold value, judging that the picture to be identified belongs to a pornographic picture; and if the lowered probability value is less than or equal to the preset second threshold value, judging that the picture to be identified does not belong to the pornographic picture.
The preset first threshold and the preset second threshold may be the same or different, for example, the preset first threshold may be set to be 40%, and the preset second threshold may be set to be 30%.
As shown in fig. 2, fig. 2 is a schematic flowchart of an embodiment of a pornographic picture identifying method according to the present invention, the pornographic picture identifying method comprising the steps of:
step S10, after receiving the picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, and the convolutional neural network model is obtained by utilizing the sample image of the preset image category to carry out training in advance.
In this embodiment, the pornographic picture identification system receives a pornographic picture identification request including a picture to be identified sent by a user, for example, the pornographic picture identification request sent by the user through a terminal such as a mobile phone, a tablet computer, or a self-service terminal device, for example, the pornographic picture identification request sent by the user from a client installed in the terminal such as the mobile phone, the tablet computer, or the self-service terminal device in advance, or the pornographic picture identification request sent by the user from a browser system in the terminal such as the mobile phone, the tablet computer, or the self-service terminal device.
After receiving a pornographic picture identification request sent by a user, the pornographic picture identification system identifies the received picture to be identified by using a pre-trained picture classification model and outputs the probability that the picture to be identified belongs to each preset picture category (pornographic picture and non-pornographic picture). The picture classification model can be trained into a model capable of effectively outputting the probability that the picture to be identified belongs to each preset picture category (pornographic picture and non-pornographic picture) by continuously performing training, learning, verification, optimization and the like by identifying a large number of preset sample pictures marked with different preset picture categories (pornographic picture and non-pornographic picture) in advance. For example, the picture classification model may employ a deep Convolutional Neural Network (CNN) model or the like.
Meanwhile, the picture format of the picture to be identified can be detected, wherein the picture format comprises a black-and-white picture and a color picture. For example, by an algorithm for judging RGB values, whether the picture to be authenticated is a black-and-white picture or a color picture is judged, and the black-and-white picture or the color picture is marked. Specifically, the manner of determining whether the picture to be identified is a black-and-white picture or a color picture is as follows: if the RGB values of the pixels in the photo are all 0 or 1, the photo to be identified is judged to be a black-and-white photo, and otherwise, the photo is a color photo.
And step S20, judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value. For example, since the pornographic pictures are generally color pictures and the probability that the pornographic pictures are black-and-white pictures is low, if the picture format of the picture to be identified is judged to be black-and-white pictures, the probability that the picture to be identified belongs to the pornographic picture category output by the picture classification model is reduced, and if the probability that the picture to be identified belongs to the pornographic picture category is still smaller than a preset threshold (such as 50%) after being reduced, the picture to be identified is judged not to belong to the pornographic pictures; if the probability that the picture to be identified belongs to the pornographic picture category is lower than a preset threshold (such as 50%), judging that the picture to be identified belongs to the pornographic picture. If the picture format of the picture to be identified is judged to be a color picture, directly comparing the probability value of the picture to be identified, which is output by the picture classification model and belongs to the pornographic picture category, with a preset threshold (such as 60%), and if the probability value is smaller than the preset threshold (such as 60%), judging that the picture to be identified does not belong to the pornographic picture; if the picture to be identified is larger than or equal to the preset threshold (such as 60%), the picture to be identified is judged to belong to the pornographic picture.
Compared with the prior art, the method and the device have the advantages that the picture to be identified is identified through the pre-trained picture classification model, and the probability that the picture to be identified belongs to each preset picture category such as pornographic pictures and non-pornographic pictures is output; detecting the picture format of the picture to be identified, such as a black-and-white picture or a color picture; and judging whether the picture to be identified belongs to the pornographic picture or not based on a preset rule according to the probability that the picture to be identified belongs to each preset picture category and the picture format of the picture to be identified. Because the probability that the pornographic picture is a black-and-white picture is lower, the method can identify the probability that the picture to be identified belongs to the pornographic picture and the non-pornographic picture, and can judge whether the picture to be identified belongs to the pornographic picture more accurately and effectively by carrying out comprehensive identification by combining the picture format of the picture to be identified such as the black-and-white picture or the color picture. Moreover, the pornographic picture identification can be automatically carried out without manual detection, and the detection efficiency is effectively improved.
In an optional embodiment, based on the above embodiment, the training step of the pre-trained image classification model is as follows:
A. setting a corresponding preset number of sample pictures for each preset picture type, and marking the corresponding preset picture type for each sample picture; the preset picture categories comprise pornographic pictures, sexy pictures and normal pictures;
B. carrying out picture pretreatment on each sample picture to obtain a training picture to be subjected to model training;
C. dividing all training pictures into a first proportion (e.g., 75%) of a training set and a second proportion of a validation set (e.g., 25%);
D. training a convolutional neural network model by using the training set;
E. and verifying the accuracy of the trained convolutional neural network model by using the verification set, if the accuracy is greater than or equal to the preset accuracy, ending the training, and taking the trained convolutional neural network model as a picture classification model, or if the accuracy is less than the preset accuracy, increasing the number of sample pictures corresponding to each preset picture category and executing the step B, C, D, E again until the accuracy of the trained convolutional neural network model is greater than or equal to the preset accuracy.
Specifically, in the present embodiment, when establishing the CNN image classification model, a large number of images of different categories are prepared, for example, 10 ten thousand sample images classified into pornographic, sexy, and normal categories are prepared, and the category to which each image belongs is labeled, for example, the category to which each image belongs may be labeled as follows, 0: the pornography; 1: the sense of sex; 2: and (4) normal. Wherein, the normal pictures can include natural, artistic and dead pictures; pictures of a sexual category may include pictures defined not to belong to pornographic pictures but between pornography and normal, such as pictorial type pictures; the pictures of the pornographic categories are defined pictures belonging to pornographic, and are the key points of picture identification in the embodiment. After the sample pictures are prepared, the sample pictures can be subjected to various pre-processing, such as being cut into training pictures with uniform size (the pictures are all resize 100X100 size) or uniform pixels. And training a Convolutional Neural Network (CNN) model with a preset model structure by using the preprocessed training pictures and the labeled categories (0: pornography, 1: sexuality and 2: normal). For example, in an alternative embodiment, the training process is as follows:
(1) establishing a convolution kernel and a deviation matrix;
(2) carrying out convolution operation by utilizing the established convolution kernel, the deviation matrix and the training picture;
(3) correcting the result of the convolution operation through the activation function of relu;
(4) pooling is a posing operation, a method of reducing an image while preserving most of the important information of the image. The pooling in this embodiment includes, but is not limited to, Mean-sampling, Max-sampling, overlaying, L2-sampling, Local Contrast Normalization, stochasticposing, Def-posing, etc.
(5) Repeating the steps (1) to (4) for 3 times;
(6) and ending the model training until the gradient of the value function of the trained model is reduced.
In addition, in the online application process of the model, if the service system finds abnormal pictures such as pornographic pictures, the found abnormal pictures can be supplemented to continue training the Convolutional Neural Network (CNN) model so as to continuously improve the identification precision of the Convolutional Neural Network (CNN) model.
After the CNN image classification model is established, the CNN image classification model can be used for image classification. The picture to be identified is input into the trained CNN picture classification model, the probability that the picture to be identified belongs to each preset picture category, namely the probability P0 that the picture to be identified belongs to the pornographic category, the probability P1 that the picture to be identified belongs to the sexual category and the probability P2 that the picture to be identified belongs to the normal category can be output through the trained CNN picture classification model.
Meanwhile, the picture format of the picture to be identified can be judged, in the embodiment, on the basis of judging that the picture to be identified is a black-and-white picture or a color picture, whether the picture to be identified is a dynamic picture such as a picture of the type of GIF (graphic interchange format) or not can be judged, and if the picture to be identified is judged to be the dynamic picture such as the type of GIF, the picture of the type of GIF is framed into a plurality of pictures for subsequent use. Specifically, there are two ways to determine whether the picture to be identified is a dynamic picture: the method comprises the steps of judging through a file suffix name of a picture to be identified, and judging through a file header of the picture to be identified in a binary format.
After the probability (P0, P1, P2) that the picture to be identified belongs to each preset picture category (pornography, sexuality, normality) is output by using the CNN picture classification model, and the format (black-white picture, color picture or dynamic picture) of the picture to be identified is judged, the probability (P0, P1, P2) and the format of the picture to be identified can be integrated, and a preset rule finally identifies whether the picture to be identified is the pornography picture. Specifically, the method comprises the following steps:
since the pornographic pictures are usually color pictures and the probability that the pornographic pictures are black-and-white pictures is low, if the picture to be identified is determined to be a black-and-white picture, the value of the probability P0 that the picture to be identified belongs to the pornographic category output by the CNN picture classification model is reduced, for example, adjusted downward by 10%, and if the reduced P0 is smaller than the first threshold, it is determined that the picture to be identified does not belong to the pornographic category, that is, P0 is 90% < 50%, and it is not determined that the picture to be identified is a pornographic picture. If the adjusted-down P0 is greater than the first threshold, the picture to be identified is determined to belong to the pornographic category, namely P0 × 90% > 50%, and the picture to be identified is determined to be the pornographic picture.
Meanwhile, the probability that the picture to be identified output by the CNN picture classification model belongs to the sexuality and normal category can be adjusted up to 10%. After the adjusted probability (P0 ', P1', P2 ') that the picture to be identified belongs to each preset picture category (pornography, sexuality and normal) is obtained, the type corresponding to MAX (P0, P1 and P2) is the identified type, the confidence rate of the type is MAX (P0, P1 and P2), that is, the picture category (pornography, sexuality or normal) corresponding to the maximum value in (P0', P1 ', P2') is the finally identified picture category to be identified. It should be noted that, on the basis that the P0 turned down in the previous step is greater than the first threshold, no matter what the picture category corresponding to the maximum value in (P0 ', P1 ', P2 ') is, the finally identified category of the picture to be identified is still a pornographic picture.
If the picture to be identified is judged to be a dynamic picture, the picture to be identified, such as a GIF (graphic interchange Format) moving picture, is cut into a plurality of pictures, the probability (P0, P1, P2) of belonging to each preset picture category (pornography, sexuality and normality) is output for each frame of the cut picture by using a CNN picture classification model, each frame of the picture is judged to belong to a black-white picture or a color picture, the pornography identification is carried out on each frame of the picture by using the rule method, and whether the picture to be identified, such as the GIF moving picture, belongs to a pornography picture or not is comprehensively judged according to the identification result of each frame of the picture to be identified, such as the GIF moving picture. For example, in an alternative embodiment, if one frame of picture is identified as a pornographic type, it is directly determined that the picture to be identified, such as the GIF moving picture, belongs to a pornographic picture. In another alternative embodiment, if both the P0' of two frames of pictures are greater than the second threshold (60% or 70%), then the picture to be authenticated, such as the GIF motion picture, is determined to belong to a pornographic picture, and so on.
The final returned result in this embodiment includes: whether the picture to be identified is a pornographic picture or not and the probability (P0 ', P1 ', P2 ') that the picture to be identified belongs to each preset picture category (pornography, sexuality and normality). In different service scenes in practical application, the first threshold value and the second threshold value can be manually set according to requirements so as to be suitable for strict or loose pornographic picture identification scenes. As in some critical scenarios, pictures belonging to a sexually sensitive category can be identified by probability (P0 ', P1 ', P2 ').
It should be noted that, when the picture formats include a black-and-white format and a non-black-and-white format, and the preset picture categories include pornographic pictures and non-pornographic pictures, the probability value that the picture to be identified belongs to the pornographic pictures and the probability value of the non-pornographic pictures are output; reducing the probability value that the picture with the black-and-white format belongs to the pornographic picture; comparing the reduced probability value with a first preset threshold value, and if the reduced probability value is greater than the preset first threshold value, judging that the picture to be identified belongs to a pornographic picture; and if the lowered probability value is less than or equal to the preset first threshold value, judging that the picture to be identified does not belong to the pornographic picture.
When the picture formats comprise a black-and-white format and a non-black-and-white format, and the preset picture categories comprise pornographic pictures, sexually sensed pictures and normal pictures, outputting the probability value of the pornographic pictures, the probability value of the sexually sensed pictures and the probability value of the normal pictures of the picture to be identified; reducing the probability value that the picture with the black-and-white format belongs to the pornographic picture; comparing the reduced probability value with a second preset threshold, and if the reduced probability value is greater than the preset second threshold, judging that the picture to be identified belongs to a pornographic picture; and if the lowered probability value is less than or equal to the preset second threshold value, judging that the picture to be identified does not belong to the pornographic picture.
The preset first threshold and the preset second threshold may be the same or different, for example, the preset first threshold may be set to be 40% and the preset second threshold may be set to be 30%.
In addition, the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a pornographic picture identification system, and the pornographic picture identification system is executable by at least one processor, so that the at least one processor executes the steps of the pornographic picture identification method in the foregoing embodiment, and the specific implementation procedures of the pornographic picture identification method, such as steps S10, S20, S30, etc., are as described above and are not repeated herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, and the scope of the invention is not limited thereby. The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments. Additionally, while a logical order is shown in the flow diagrams, in some cases, the steps shown or described may be performed in an order different than here.
Those skilled in the art can implement the invention in various modifications, such as features from one embodiment can be used in another embodiment to yield yet a further embodiment, without departing from the scope and spirit of the invention. Any modification, equivalent replacement and improvement made within the technical idea of using the present invention should be within the scope of the right of the present invention.
Claims (7)
1. A pornographic picture identification method is characterized by comprising the following steps:
after receiving a picture to be identified, detecting the picture format of the picture to be identified, identifying the picture to be identified by utilizing a pre-trained picture classification model, and outputting the probability value that the picture to be identified belongs to one or more preset picture categories; the image classification model is a convolutional neural network model, the convolutional neural network model is obtained by utilizing sample images of preset image categories to train in advance, the image formats comprise a black-and-white format and a non-black-and-white format, and the preset image categories comprise pornographic images and non-pornographic images;
judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format of the picture to be identified and the probability value, wherein the judging step comprises the following steps of: if the picture format of the picture to be identified is judged to be a black and white picture, the probability value of the picture to be identified, which is output by the picture classification model and belongs to the pornographic picture category, is reduced by a preset value, if the probability value of the picture to be identified, which belongs to the pornographic picture category, is reduced by the preset value and is still smaller than a preset threshold value, the picture to be identified is judged not to belong to the pornographic picture, and if the probability value of the picture to be identified, which belongs to the pornographic picture category, is reduced by the preset value and is larger than the preset threshold value, the picture to be identified is judged to belong to the pornographic picture.
2. The pornographic picture identification method according to claim 1, wherein the picture classification model is trained by the steps of:
A. collecting a preset number of pornographic pictures, sexy pictures and normal pictures as sample pictures, and marking corresponding picture categories on each sample picture;
B. preprocessing each sample picture;
C. dividing the preprocessed sample picture into a training set with a first proportion and a verification set with a second proportion;
D. training a convolutional neural network model by using the training set;
E. verifying the accuracy of the trained convolutional neural network model by using the verification set, finishing the training if the accuracy is greater than or equal to a preset accuracy, and taking the trained convolutional neural network model as a picture classification model; if the accuracy is less than the predetermined accuracy, the sample picture is collected again and the above step B, C, D, E is repeated.
3. The pornographic picture identification method according to claim 2, wherein the picture classification model is trained as follows:
H. establishing a convolution kernel and a deviation matrix;
I. carrying out convolution operation by utilizing the established convolution kernel, the deviation matrix and a sample picture;
J. correcting the result of the convolution operation through an activation function of relu;
K. and performing pooling operation on the sample pictures, and repeatedly executing the step H, I, J until the gradient of the value function of the trained model is reduced, thereby finishing the model training.
4. The pornographic picture identification method according to claim 1, wherein when the picture to be identified is a moving picture, further comprising the steps of:
the picture to be identified is cut into a plurality of pictures, the picture format of each picture of the cut frames is detected, the plurality of pictures of the cut frames are respectively identified by utilizing a pre-trained picture classification model, and the probability value of each picture of the cut frames belonging to each preset picture category is output;
and judging whether the picture to be identified belongs to the pornographic picture or not according to the picture format and the probability value of each picture of the cut frame.
5. The method according to claim 4, wherein the step of determining whether the picture to be evaluated belongs to the pornographic picture according to the picture format and the probability value of each picture of the cut frame specifically comprises:
judging whether pictures belonging to pornographic pictures exist in the cut-frame pictures or not according to the picture format and the probability value of each picture of the cut-frame;
if the picture to be identified belongs to the pornographic picture, judging that the picture to be identified belongs to the pornographic picture; otherwise, judging that the picture to be identified does not belong to the pornographic picture; or
Judging whether more than two pictures with pornographic picture probability values larger than a preset value in the pictures of the cut frames exist according to the picture formats of the pictures of the cut frames and the probability values;
if so, judging that the picture to be identified belongs to the pornographic picture; otherwise, judging that the picture to be identified does not belong to the pornographic picture.
6. An electronic device, comprising a memory, a processor, the memory having stored thereon a pornographic picture evaluation system executable on the processor, the pornographic picture evaluation system when executed by the processor implementing the steps of the pornographic picture evaluation method according to any one of claims 1-5.
7. A computer-readable storage medium, having stored thereon a pornographic picture identification system, the pornographic picture identification system, when executed by a processor, implementing the steps of the pornographic picture identification method according to any one of claims 1-5.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810184583.6A CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
| PCT/CN2018/089718 WO2019169767A1 (en) | 2018-03-06 | 2018-06-03 | Pornographic picture identification method, electronic device, and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810184583.6A CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108491866A CN108491866A (en) | 2018-09-04 |
| CN108491866B true CN108491866B (en) | 2022-09-13 |
Family
ID=63341693
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810184583.6A Active CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108491866B (en) |
| WO (1) | WO2019169767A1 (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109491970A (en) * | 2018-10-11 | 2019-03-19 | 平安科技（深圳）有限公司 | Imperfect picture detection method, device and storage medium towards cloud storage |
| CN109523281A (en) * | 2018-11-26 | 2019-03-26 | 北京拓世寰宇网络技术有限公司 | A kind of determining source of houses picture category method for distinguishing and device |
| CN109886865B (en) * | 2019-01-07 | 2024-01-23 | 平安科技（深圳）有限公司 | Method, device, computer equipment and storage medium for automatically shielding bad information |
| CN110210356A (en) * | 2019-05-24 | 2019-09-06 | 厦门美柚信息科技有限公司 | A kind of picture discrimination method, apparatus and system |
| CN110456955B (en) * | 2019-08-01 | 2022-03-29 | 腾讯科技（深圳）有限公司 | Exposed clothing detection method, device, system, equipment and storage medium |
| CN111199233B (en) * | 2019-12-30 | 2020-11-20 | 四川大学 | Improved deep learning pornographic image identification method |
| CN111259823A (en) * | 2020-01-19 | 2020-06-09 | 人民中科(山东)智能技术有限公司 | Pornographic image identification method based on convolutional neural network |
| CN115546824B (en) * | 2022-04-18 | 2023-11-28 | 荣耀终端有限公司 | Taboo picture identification method, apparatus and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105761201A (en) * | 2016-02-02 | 2016-07-13 | 山东大学 | Method for translation of characters in picture |
| CN106874924A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of picture style and device |
| CN107067020A (en) * | 2016-12-30 | 2017-08-18 | 腾讯科技（上海）有限公司 | Image identification method and device |
| CN107180225A (en) * | 2017-04-19 | 2017-09-19 | 华南理工大学 | A kind of recognition methods for cartoon figure's facial expression |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| HUE048285T2 (en) * | 2012-01-11 | 2020-07-28 | 77 Elektronika Mueszeripari Kft | Two stage categorization of objects in images |
| CN105989330A (en) * | 2015-02-03 | 2016-10-05 | 阿里巴巴集团控股有限公司 | Picture detection method and apparatus |
| CN106874296B (en) * | 2015-12-14 | 2021-06-04 | 阿里巴巴集团控股有限公司 | Method and device for identifying style of commodity |
| CN106096602A (en) * | 2016-06-21 | 2016-11-09 | 苏州大学 | A kind of Chinese licence plate recognition method based on convolutional neural networks |
| US10083369B2 (en) * | 2016-07-01 | 2018-09-25 | Ricoh Company, Ltd. | Active view planning by deep learning |
| CN106228158A (en) * | 2016-07-25 | 2016-12-14 | 北京小米移动软件有限公司 | The method and apparatus of picture detection |
| CN106951848B (en) * | 2017-03-13 | 2018-12-28 | 平安科技（深圳）有限公司 | The method and system of picture recognition |
| CN107330453B (en) * | 2017-06-19 | 2020-07-07 | 中国传媒大学 | Pornographic image identification method based on step-by-step identification and fusion key part detection |
| CN107665333A (en) * | 2017-08-28 | 2018-02-06 | 平安科技（深圳）有限公司 | A kind of indecency image identification method, terminal, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107665353A (en) * | 2017-09-15 | 2018-02-06 | 平安科技（深圳）有限公司 | Model recognizing method, device, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107729924B (en) * | 2017-09-25 | 2019-02-19 | 平安科技（深圳）有限公司 | Picture review probability interval generation method and picture review determination method |
-
2018
- 2018-03-06 CN CN201810184583.6A patent/CN108491866B/en active Active
- 2018-06-03 WO PCT/CN2018/089718 patent/WO2019169767A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106874924A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of picture style and device |
| CN105761201A (en) * | 2016-02-02 | 2016-07-13 | 山东大学 | Method for translation of characters in picture |
| CN107067020A (en) * | 2016-12-30 | 2017-08-18 | 腾讯科技（上海）有限公司 | Image identification method and device |
| CN107180225A (en) * | 2017-04-19 | 2017-09-19 | 华南理工大学 | A kind of recognition methods for cartoon figure's facial expression |
Non-Patent Citations (1)
| Title |
|---|
| 基于纠错编码的CSNN及其在遥感图像分类中的应用;蒋艳凰等;《计算机研究与发展》;20030730(第07期);20-26 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108491866A (en) | 2018-09-04 |
| WO2019169767A1 (en) | 2019-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108491866B (en) | Pornographic picture identification method, electronic device and readable storage medium | |
| WO2019174130A1 (en) | Bill recognition method, server, and computer readable storage medium | |
| WO2019109526A1 (en) | Method and device for age recognition of face image, storage medium | |
| CN109829453B (en) | Method and device for recognizing characters in card and computing equipment | |
| CN112861648B (en) | Character recognition method, character recognition device, electronic equipment and storage medium | |
| WO2018010657A1 (en) | Structured text detection method and system, and computing device | |
| CN110555372A (en) | Data entry method, device, equipment and storage medium | |
| WO2018166116A1 (en) | Car damage recognition method, electronic apparatus and computer-readable storage medium | |
| CN110033018B (en) | Graph similarity judging method and device and computer readable storage medium | |
| US10402640B1 (en) | Method and system for schematizing fields in documents | |
| CN110781460A (en) | Copyright authentication method, device, equipment, system and computer readable storage medium | |
| CN110008997B (en) | Image texture similarity recognition method, device and computer readable storage medium | |
| CN111612081B (en) | Training method, device, equipment and storage medium for recognition model | |
| WO2019062081A1 (en) | Salesman profile formation method, electronic device and computer readable storage medium | |
| CN114881698A (en) | Advertisement compliance auditing method and device, electronic equipment and storage medium | |
| CN113111880A (en) | Certificate image correction method and device, electronic equipment and storage medium | |
| JP6882362B2 (en) | Systems and methods for identifying images, including identification documents | |
| CN108268778B (en) | Data processing method, device and storage medium | |
| CN114386013A (en) | Automatic student status authentication method and device, computer equipment and storage medium | |
| CN107992872B (en) | Method for carrying out text recognition on picture and mobile terminal | |
| CN111368709A (en) | Picture text recognition method, device and equipment and readable storage medium | |
| CN111163332A (en) | Video pornography detection method, terminal and medium | |
| CN113239910B (en) | Certificate identification method, device, equipment and storage medium | |
| CN115063826A (en) | Mobile terminal driver license identification method and system based on deep learning | |
| CN113780116A (en) | Invoice classification method and device, computer equipment 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 |