CN110781939A - Method and device for detecting similar pictures and project management system - Google Patents
Method and device for detecting similar pictures and project management system Download PDFInfo
- Publication number
- CN110781939A CN110781939A CN201910989869.6A CN201910989869A CN110781939A CN 110781939 A CN110781939 A CN 110781939A CN 201910989869 A CN201910989869 A CN 201910989869A CN 110781939 A CN110781939 A CN 110781939A
- Authority
- CN
- China
- Prior art keywords
- picture
- pictures
- detected
- item
- similar pictures
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- 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
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses a method and a device for detecting similar pictures and a project management system, wherein the method comprises the following steps: acquiring local context information of each picture in a picture set of a project to be detected and global feature information of each picture in the picture set of the project to be detected; based on the bidirectional long-time and short-time memory neural network model BLSTM and the full-connection neural network model, similar pictures with similarity within a preset range in a to-be-detected project picture set are detected and obtained through local context information of each picture and global feature information of each picture, classification results of the similar pictures are obtained, the detection results represent the similarity condition of the project pictures, the method can be used for managing the project construction pictures, labor cost is reduced, management efficiency is improved, and the condition that a user replaces the pictures in the current link with the pictures in other projects or other processes, which are not in accordance with the specifications, is avoided.
Description
Technical Field
The invention relates to the technical field of computer information management, in particular to a method and a device for detecting similar pictures and a project management system.
Background
With the vigorous development of the information-oriented society, the scientific technology is changed day by day, the intelligent technology is gradually and widely applied to various fields, and meanwhile, the information-oriented construction of the enterprise system is deeply influenced. As the leading-edge technology in the field of computers at present, the artificial intelligence technology can solve a plurality of practical problems by learning the thinking of human brains, and the development trend is very strong. Among them, the image detection and identification technology is undoubtedly an important branch under artificial intelligence.
At present, an enterprise generally manages a project full life cycle through a Project Management System (PMS), and in a project implementation process, a user reports a project implementation progress situation in real time through a mobile phone APP and a computer terminal and uploads process pictures of implementation procedures in batches. However, in the project implementation process, not only different process procedures are required for each product type, but also the process procedures of different projects are different, and the number of projects under construction of the system is huge, and these factors may cause that when a user uploads the pictures of the project construction site, in order to speed up the progress of the on-line project construction, the situation that the pictures of the current link do not meet the specification is replaced by the existing pictures of other projects or other processes occurs, and in the project management process, it is very inconvenient to manually screen and check such irregular pictures, the labor cost is high, and the management is not convenient.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method and a device for detecting similar pictures and a project management system, and solves the problems of high labor cost and inconvenience in management due to the fact that the repeated situation of construction pictures of projects uploaded by users is checked manually in the project management process.
According to an aspect of the present invention, a method for detecting similar pictures is provided, including:
acquiring local context information of each picture in a picture set of the item to be detected;
acquiring global feature information of each picture in a picture set of the item to be detected;
based on a bidirectional Long Short-term memory neural network model (BLSTM) and a full-connection neural network model, detecting similar pictures with similarity in a preset range in a picture set of the item to be detected according to local context information of each picture and global feature information of each picture, and obtaining classification results of the similar pictures.
Optionally, the obtaining local context information of each picture in the picture set to be detected includes:
acquiring a detection window set corresponding to each picture in a picture set of the item to be detected; the detection window set comprises a plurality of windows, and each window stores window characteristic information;
and constructing an object relation graph of each picture according to the window characteristic information, and taking the object relation graph as local context information of each picture.
Optionally, the window characteristic information includes at least one of:
label information of similar picture categories to which the window belongs;
probability scores corresponding to the similar picture categories;
the abscissa and ordinate of the center of the window;
the width and height of the window.
Optionally, the obtaining global feature information of each picture in the picture set of the item to be detected includes:
and filtering each picture through a Gabor filter to obtain the contour information of the object in each picture, and taking the contour information of the object as the global feature information of each picture.
Optionally, based on the bidirectional long-and-short-term memory neural network model BLSTM and the fully-connected neural network model, according to the local context information of each picture and the global feature information of each picture, detecting similar pictures with similarity within a preset range in the picture set of the item to be detected, and obtaining classification results of the similar pictures, including:
inputting the local context information of each picture and the global feature information of each picture into the BLSTM, and outputting the global context feature of each picture;
and jointly inputting the global context characteristics of all pictures in the item picture set to be detected into the fully-connected neural network model, detecting similar pictures with the similarity within a preset range in the item picture set to be detected, and outputting the classification results of the similar pictures.
Optionally, based on the bidirectional long-and-short-term memory neural network model BLSTM and the fully-connected neural network model, according to the local context information of each picture and the global feature information of each picture, similar pictures with similarity within a preset range in the item picture set to be detected, and after obtaining the classification result of the similar pictures, the method further includes:
and obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures.
Optionally, obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures includes:
analyzing and calculating the classification result of the similar pictures, acquiring the proportion of the similar pictures in the total number of the pictures of the item to be detected, and acquiring similar pictures with the number of completely repeated pictures and the number of highly similar pictures;
and/or the presence of a gas in the gas,
and analyzing and calculating the classification result of the similar pictures, and acquiring the number of projects and the number of pictures which use the same pictures across the projects.
According to another aspect of the present invention, there is provided a device for detecting similar pictures, comprising:
the first acquisition module is used for acquiring local context information of each picture in the picture set of the item to be detected; and
the second acquisition module is used for acquiring the global characteristic information of each picture in the picture set of the item to be detected;
and the detection module is used for detecting similar pictures with the similarity within a preset range in the picture set of the item to be detected based on the bidirectional long-short time memory neural network model BLSTM and the full-connection neural network model according to the local context information of each picture and the global feature information of each picture, and obtaining the classification results of the similar pictures.
According to another aspect of the present invention, there is provided a project management system, the system including a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the method for detecting similar pictures as described above when executing the computer program.
According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when being executed by a processor, implements the steps of the method for detecting a similar picture as described above.
The embodiment of the invention has the beneficial effects that:
in the scheme, the local context information of each picture in the picture set of the item to be detected and the global feature information of each picture in the picture set of the item to be detected are obtained; based on the bidirectional long-short time memory neural network model BLSTM and the full-connection neural network model, similar pictures with similarity within a preset range in the picture set of the item to be detected are detected and obtained through the local context information of each picture and the global feature information of each picture, the classification results of the similar pictures are obtained, the detection results represent the similarity condition of the item pictures, the labor cost of checking construction site pictures of similar items uploaded by users in the process of item management is reduced, and the efficiency and the normalization of the item management are improved.
Drawings
FIG. 1 is a flowchart illustrating a method for detecting similar pictures according to an embodiment of the present invention;
FIG. 2 is a diagram of a similar picture detection architecture according to an embodiment of the present invention;
FIG. 3 is a second flowchart of a similar picture detection method according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating an embodiment of obtaining a picture repetition status in a picture set of an item to be detected;
FIG. 5 is a third flowchart illustrating a method for detecting similar pictures according to an embodiment of the present invention;
fig. 6 is a block diagram of a similar picture detection apparatus according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
As shown in fig. 1, an embodiment of the present invention provides a method for detecting similar pictures, including:
and step 11, acquiring local context information of each picture in the picture set of the item to be detected.
Here, step 11 may further include, before, acquiring a picture set of the item to be detected. Specifically, in an application scenario where the project management system performs project management, in order to detect a repetition of project pictures uploaded by related personnel during a project implementation process of all projects in a certain area, a picture detection task may be first created in the system, and the detection task records and displays information such as task creation time, start time, end time, and task state. And then sequentially reading undetected tasks from the newly-built picture detection task queue through a background timer, acquiring picture information contained in all project processes in the area to which the task belongs every time a new task is read, and taking the picture information contained in all project processes in the area to which the task belongs as the picture set of the project to be detected.
And step 12, acquiring global characteristic information of each picture in the picture set of the item to be detected.
In this embodiment, the global feature information refers to a global Gist feature. The Gist characteristic is a global characteristic descriptor, which reflects the interaction between an object and a scene, for example, a certain association relationship is often formed between an iron tower and a tree, between a lamp post and an electricity meter box, and the like in a construction picture, which is equivalent to the spatial representation of an image. The Gist characteristic of each picture is combined to improve the accuracy of subsequent identification and classification of similar pictures.
It should be noted that each picture may be filtered through a Gabor filter to obtain the contour information of the object in each picture, and the contour information of the object is used as the global Gist feature of each picture. In detail, the contour information of the object in the picture can be obtained and extracted as the global Gist characteristic of each picture after filtering operation is performed on each picture by using a Gabor filter bank.
And step 13, detecting similar pictures with similarity within a preset range in the picture set of the item to be detected based on the bidirectional long-short time memory neural network model BLSTM and the full-connection neural network model according to the local context information of each picture and the global feature information of each picture, and obtaining the classification results of the similar pictures.
In the embodiment, the long-distance memory neural network LSTM unit effectively solves the long-distance dependence problem and updates the storage state by using a gate control mechanism, and history information is deleted, reserved or updated in the next unit by using an input gate, a forgetting gate, an output gate and selection. The bidirectional long-time and short-time memory neural network model BLSTM is added with a hidden layer. As shown in FIG. 2, for the architecture diagram of similar picture detection, in FIG. 2, the input of the input layer includes two parts, one part is the local context information (X) of the project construction picture
1,X
2,X
3，X
4) And one part is the global feature information (global Gist feature) of the system, and the system are spliced into the BLSTM layer.
A fully connected layer (FC for short), i.e. a fully connected neural network, is a most basic neural network structure, and uses a state transition matrix as a parameter to optimize and update the parameter according to a calculation loss function and a calculation result of random gradient descent. The role of the fully-connected neural network model is to perform final classification on similar pictures according to the output (y1, y2, y3, y4) of the BLSTM layer, by which context window information can be effectively used to predict whether or not windows of a current picture belong to similar categories and output the same.
In an alternative embodiment of the present invention, step 11 includes:
acquiring a detection window set corresponding to each picture in a picture set of the item to be detected; the detection window set comprises a plurality of windows, and each window stores window characteristic information;
wherein the window characteristic information may include at least one of: label information of similar picture categories to which the window belongs; probability scores corresponding to the similar picture categories; the abscissa and ordinate of the center of the window; the width and height of the window.
And constructing an object relation graph of each picture according to the window characteristic information, and taking the object relation graph as local context information of each picture.
In this embodiment, for each picture, horizontal or vertical detection is performed on the picture according to a certain rule in a window sliding manner until the whole image is traversed, so as to obtain M window sets, where M is a positive integer. Each window stores information such as: obtaining a feature vector set { X, X) of M windows by combining the label information Pm of the similar picture category to which the window belongs, the probability score Gm corresponding to the category and output by the object classifier, the abscissa and ordinate of the center of the window and the width and height of the window, and the information stored in each window
1,X
2,...,X
MWhere M is a positive integer. And further constructing an object relation graph of each picture according to semantic context, spatial angle context and the like among the M windows in each picture, wherein the object relation graph is used as local context information of each picture.
In an optional embodiment of the present invention, step 13 includes:
and inputting the local context information of each picture and the global feature information of each picture into the BLSTM, and outputting the global context feature of each picture.
In this embodiment, BLSTM adds a hidden layer to simultaneously contact information for any window and its forward part in a set of windows in an image, as shown in fig. 2
And contact information of backward part
Performing bidirectional grabbing to form the global context feature (y) of the picture
1，y
2，y
1，y
1)。
In an optional embodiment of the present invention, step 13 further includes:
and jointly inputting the global context characteristics of all pictures in the item picture set to be detected into the fully-connected neural network model, detecting similar pictures with the similarity within a preset range in the item picture set to be detected, and outputting the classification results of the similar pictures.
In the embodiment, the global context feature information (output of the BLSTM layer) of each picture is jointly input and connected by using a full connection layer (a full connection neural network model), and an optimal solution is obtained through a Softmax regression function. And the full-connection layer uses the state transition matrix as a parameter, and optimizes and updates the parameter according to a calculation loss function and a calculation result of random gradient descent. By the layer, the context window information can be effectively used for predicting whether each window of the current picture belongs to a similar category or not and outputting the window.
As shown in fig. 3, in an alternative embodiment of the present invention, after step 13, the method further includes:
and 14, obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures.
In the embodiment, the picture repetition condition in the picture set of the item to be detected is obtained, so that the picture repetition condition of the item is displayed more intuitively, and the item management is facilitated.
Specifically, step 14 may include:
analyzing and calculating the classification result of the similar pictures, acquiring the proportion of the similar pictures in the total number of the pictures of the item to be detected, and acquiring similar pictures with the number of completely repeated pictures and the number of highly similar pictures;
and/or the presence of a gas in the gas,
and analyzing and calculating the classification result of the similar pictures, and acquiring the number of projects and the number of pictures which use the same pictures across the projects.
In the embodiment, the user can optionally obtain detailed information of the project picture repetition condition from the picture dimension or the project dimension or the two dimensions of the picture and the project according to needs, check the normalization of the project picture uploading, facilitate self-checking and other checking, and improve the project management efficiency.
As shown in fig. 4, it shows an item picture repetition case acquisition flow.
In fig. 4, in terms of picture dimensions, the number of similar pictures is easily obtained according to the classification result of the similar pictures (picture detection and comparison result in the picture), then the total number of the pictures can be obtained by searching the accessory picture resource library (picture library), the system automatically traverses and calculates the proportion of the similar pictures in all the pictures, and counts the number of completely repeated pictures and the number of highly similar pictures;
in fig. 4, in terms of the dimension of the item, by loading the item information stream (e.g., proctoring unit, construction unit, item name, process name, etc.) carried by the picture itself and combining the total number of items recorded in the system, it is checked whether each picture in each similar picture group (classification of similar pictures) belongs to a different item, and the number of the pictures is accumulated at the same time, so that the number of items and the number of pictures using the same picture across the items can be obtained.
Further, optionally, the similar pictures and the project data information can be collected according to the three-level organization of the whole country, so that the uploading quality of the project pictures of the whole management country-province-city is realized, meanwhile, a user can obtain the detailed information of the similar pictures according to the result of each dimension, and whether the project construction process meets the standard or not is self-checked, so that the correct and qualified process pictures can be uploaded and replaced again in time; the inspection personnel can also check the project according to the detection result.
Furthermore, the repeated condition of the pictures can be displayed by utilizing platforms such as a project management system and the like. Further, the presenting content may include: a data display area and a picture display area.
An implementation flow chart of the similar picture detection is described below with reference to fig. 5. As shown in fig. 5, the method mainly comprises the following steps:
And step 52, acquiring project pictures. And the background timer sequentially reads undetected tasks from the newly-built task queue, and when a new task is read, firstly, a picture set included in all project procedures of the city to which the task belongs is obtained, wherein the picture set is a project picture set to be detected.
And step 53, carrying out image recognition detection on the item picture set. The method comprises the steps of inputting local context information of each project picture, extracting global context characteristics by using a BLTSM neural network model, and finally classifying the pictures by using a full-connection network layer, wherein the pictures with the similarity exceeding a certain threshold are classified into one class and used as a similar picture detection result (referring to the classification result of the similar pictures).
And step 54, analyzing and calculating the output picture result set according to the similar picture detection result, and displaying the repeated condition of the project picture more intuitively.
The similar picture detection method is applied to the application scenes including but not limited to the inspection of projects by inspection mechanisms such as self-inspection and audit of project units in the project management process. According to the embodiment, the manpower and time of the rechecking link are saved, and the engineering construction efficiency is improved.
The invention also provides a device for realizing the method.
As shown in fig. 6, the present invention provides a device for detecting similar pictures, where the device 600 includes:
a first obtaining module 601, configured to obtain local context information of each picture in a picture set of an item to be detected; and
a second obtaining module 602, configured to obtain global feature information of each picture in the picture set of the item to be detected;
the detection module 603 is configured to detect similar pictures with similarity within a preset range in the picture set of the item to be detected based on the bidirectional long-and-short-term memory neural network model BLSTM and the fully-connected neural network model according to the local context information of each picture and the global feature information of each picture, and obtain a classification result of the similar pictures.
In an alternative embodiment of the present invention, the first obtaining module 601 includes:
the first acquisition unit is used for acquiring a detection window set corresponding to each picture in the picture set of the item to be detected; the detection window set comprises a plurality of windows, and each window stores window characteristic information;
and the second acquisition unit is used for constructing an object relation graph of each picture according to the window characteristic information and taking the object relation graph as local context information of each picture.
Optionally, the window characteristic information includes at least one of:
label information of similar picture categories to which the window belongs;
probability scores corresponding to the similar picture categories;
the abscissa and ordinate of the center of the window;
the width and height of the window.
In an optional embodiment of the present invention, the second obtaining module 602 is specifically configured to:
and filtering each picture through a Gabor filter to obtain the contour information of the object in each picture, and taking the contour information of the object as the global feature information of each picture.
In an optional embodiment of the present invention, the detecting module 603 includes:
the first detection unit is used for inputting the local context information of each picture and the global feature information of each picture into the BLSTM and outputting the global context feature of each picture;
and the second detection unit is used for jointly inputting the global context characteristics of all the pictures in the item picture set to be detected into the fully-connected neural network model, detecting similar pictures with the similarity within a preset range in the item picture set to be detected, and outputting the classification results of the similar pictures.
In an optional embodiment of the present invention, the apparatus 600 further includes:
and the processing module is used for obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures.
Specifically, the processing module may include:
the first processing unit is used for analyzing and calculating the classification result of the similar pictures, acquiring the proportion of the similar pictures in the total number of the pictures of the item to be detected, and acquiring the similar pictures with the completely repeated pictures and the highly similar pictures;
and/or the presence of a gas in the gas,
and the second processing unit is used for analyzing and calculating the classification result of the similar pictures and acquiring the number of items and the number of pictures which use the same pictures across the items.
The device is a device corresponding to the method embodiment, and all implementation manners in the method embodiment are applicable to the device embodiment, and the same technical effects as the method embodiment can be achieved.
The invention also provides a project management system, which comprises a processor, a memory and a computer program stored on the memory and capable of running on the processor, wherein the processor implements the steps of the method for detecting the similar pictures when executing the computer program.
The present invention also provides a computer-readable storage medium having stored thereon a computer program, which, when being executed by a processor, implements the steps of the method for detecting a similar picture as described above.
According to the scheme, the attachment uploading function of the pictures in the project construction process is combined with the picture identification technology, the neural network architecture model combining the BLTSM (bidirectional long and short term memory neural network) and the full-connection neural network is utilized to identify and classify the project construction pictures one by one, the local context characteristics and the global context characteristics of the pictures are combined and sent to the BLSTM layer for training, the full-connection layer is used for classification mapping, and the good identification effect is achieved. Finally, the completely same or highly similar picture sets are obtained and displayed in the system, and an informatization system, a service scene and an advanced technology are fully and effectively combined together, so that the availability and the convenience of the system are improved, a solid foundation is laid for creating an intelligent platform for project management, a user can know the implementation condition of a construction project more easily, and self-check or examination can be performed in time; meanwhile, the pictures are automatically compared through the program, so that the manpower and time of a rechecking link are saved, and the engineering construction efficiency is improved.
While the preferred embodiments of the present invention have been described, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the following claims.
Claims (10)
1. A method for detecting similar pictures is characterized by comprising the following steps:
acquiring local context information of each picture in a picture set of the item to be detected;
acquiring global feature information of each picture in a picture set of the item to be detected;
based on the bidirectional long-short time memory neural network model BLSTM and the full-connection neural network model, detecting similar pictures with similarity within a preset range in the picture set of the item to be detected according to the local context information of each picture and the global feature information of each picture, and obtaining the classification results of the similar pictures.
2. The method for detecting similar pictures according to claim 1, wherein obtaining local context information of each picture in the set of pictures to be detected comprises:
acquiring a detection window set corresponding to each picture in a picture set of the item to be detected; the detection window set comprises a plurality of windows, and each window stores window characteristic information;
and constructing an object relation graph of each picture according to the window characteristic information, and taking the object relation graph as local context information of each picture.
3. The method according to claim 2, wherein the window characteristic information includes at least one of:
label information of similar picture categories to which the window belongs;
probability scores corresponding to the similar picture categories;
the abscissa and ordinate of the center of the window;
the width and height of the window.
4. The method for detecting similar pictures according to claim 1, wherein the obtaining of global feature information of each picture in the picture set of the item to be detected comprises:
and filtering each picture through a Gabor filter to obtain the contour information of the object in each picture, and taking the contour information of the object as the global feature information of each picture.
5. The method for detecting similar pictures according to claim 1, wherein the method for detecting similar pictures with similarity in a preset range in a picture set of items to be detected based on a bidirectional long-and-short time memory neural network model BLSTM and a full-connection neural network model according to local context information of each picture and global feature information of each picture and obtaining classification results of the similar pictures comprises:
inputting the local context information of each picture and the global feature information of each picture into the BLSTM, and outputting the global context feature of each picture;
and jointly inputting the global context characteristics of all pictures in the item picture set to be detected into the fully-connected neural network model, detecting similar pictures with the similarity within a preset range in the item picture set to be detected, and outputting the classification results of the similar pictures.
6. The method for detecting similar pictures according to claim 1, wherein based on the bi-directional long-short time memory neural network model BLSTM and the fully connected neural network model, according to the local context information of each picture and the global feature information of each picture, the method further comprises the steps of, after obtaining the classification result of the similar pictures, detecting similar pictures with similarity in a preset range in the item picture set to be detected:
and obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures.
7. The method for detecting similar pictures according to claim 6, wherein obtaining the picture repetition condition in the picture set of the item to be detected according to the classification result of the similar pictures comprises:
analyzing and calculating the classification result of the similar pictures, acquiring the proportion of the similar pictures in the total number of the pictures of the item to be detected, and acquiring similar pictures with the number of completely repeated pictures and the number of highly similar pictures;
and/or the presence of a gas in the gas,
and analyzing and calculating the classification result of the similar pictures, and acquiring the number of projects and the number of pictures which use the same pictures across the projects.
8. An apparatus for detecting a similar picture, comprising:
the first acquisition module is used for acquiring local context information of each picture in the picture set of the item to be detected; and
the second acquisition module is used for acquiring the global characteristic information of each picture in the picture set of the item to be detected;
and the detection module is used for detecting similar pictures with the similarity within a preset range in the picture set of the item to be detected based on the bidirectional long-short time memory neural network model BLSTM and the full-connection neural network model according to the local context information of each picture and the global feature information of each picture, and obtaining the classification results of the similar pictures.
9. An item management system, characterized in that the system comprises a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method for detecting similar pictures according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for detecting a similar picture according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910989869.6A CN110781939A (en) | 2019-10-17 | 2019-10-17 | Method and device for detecting similar pictures and project management system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910989869.6A CN110781939A (en) | 2019-10-17 | 2019-10-17 | Method and device for detecting similar pictures and project management system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110781939A true CN110781939A (en) | 2020-02-11 |
Family
ID=69385870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910989869.6A Pending CN110781939A (en) | 2019-10-17 | 2019-10-17 | Method and device for detecting similar pictures and project management system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110781939A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169035A (en) * | 2017-04-19 | 2017-09-15 | 华南理工大学 | A kind of file classification method for mixing shot and long term memory network and convolutional neural networks |
| CN108960127A (en) * | 2018-06-29 | 2018-12-07 | 厦门大学 | Pedestrian's recognition methods again is blocked based on the study of adaptive depth measure |
| CN109446897A (en) * | 2018-09-19 | 2019-03-08 | 清华大学 | Scene recognition method and device based on image context information |
| US20190122103A1 (en) * | 2017-10-24 | 2019-04-25 | International Business Machines Corporation | Attention based sequential image processing |
| CN110196906A (en) * | 2019-01-04 | 2019-09-03 | 华南理工大学 | Towards financial industry based on deep learning text similarity detection method |
-
2019
- 2019-10-17 CN CN201910989869.6A patent/CN110781939A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169035A (en) * | 2017-04-19 | 2017-09-15 | 华南理工大学 | A kind of file classification method for mixing shot and long term memory network and convolutional neural networks |
| US20190122103A1 (en) * | 2017-10-24 | 2019-04-25 | International Business Machines Corporation | Attention based sequential image processing |
| CN108960127A (en) * | 2018-06-29 | 2018-12-07 | 厦门大学 | Pedestrian's recognition methods again is blocked based on the study of adaptive depth measure |
| CN109446897A (en) * | 2018-09-19 | 2019-03-08 | 清华大学 | Scene recognition method and device based on image context information |
| CN110196906A (en) * | 2019-01-04 | 2019-09-03 | 华南理工大学 | Towards financial industry based on deep learning text similarity detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112861975B (en) | Classification model generation method, classification device, electronic equipment and medium | |
| CN112632385A (en) | Course recommendation method and device, computer equipment and medium | |
| CN108804704A (en) | A kind of user's depth portrait method and device | |
| CN103348315B (en) | Content storage management in video camera | |
| CN112613569B (en) | Image recognition method, training method and device for image classification model | |
| CN112070135A (en) | Power equipment image detection method and device, power equipment and storage medium | |
| CN107229731A (en) | Method and apparatus for grouped data | |
| CN115131283A (en) | Defect detection and model training method, device, equipment and medium for target object | |
| CN114550051A (en) | Vehicle loss detection method and device, computer equipment and storage medium | |
| CN109857878B (en) | Article labeling method and device, electronic equipment and storage medium | |
| Bhargava et al. | A study on potential of big visual data analytics in construction Arena | |
| CN114419631A (en) | Network management virtual system based on RPA | |
| CN110276405A (en) | Method and apparatus for output information | |
| CN112597238A (en) | Method, system, device and medium for establishing knowledge graph based on personnel information | |
| CN113139452A (en) | Method for detecting behavior of using mobile phone based on target detection | |
| CN110781939A (en) | Method and device for detecting similar pictures and project management system | |
| CN110689028A (en) | Site map evaluation method, site survey record evaluation method and site survey record evaluation device | |
| CN114693435A (en) | Intelligent return visit method and device for collection list, electronic equipment and storage medium | |
| CN111126629B (en) | Model generation method, brush list identification method, system, equipment and medium | |
| Sun et al. | Automatic building age prediction from street view images | |
| CN114299295A (en) | Data processing method and related device | |
| CN112347889A (en) | Substation operation behavior identification method and device | |
| CN112256836A (en) | Recording data processing method and device and server | |
| CN112651996A (en) | Target detection tracking method and device, electronic equipment and storage medium | |
| CN113723093B (en) | Personnel management policy recommendation 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 | ||
| CB02 | Change of applicant information |
Address after: Room 101, floors 1-3, building 14, North District, yard 9, dongran North Street, Haidian District, Beijing 100029 Applicant after: CHINA TOWER Co.,Ltd. Address before: 100142 19th floor, 73 Fucheng Road, Haidian District, Beijing Applicant before: CHINA TOWER Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200211 |
|
| RJ01 | Rejection of invention patent application after publication |