WO2019148923A1 - 一种以图搜图方法、装置、电子设备及存储介质 - Google Patents
一种以图搜图方法、装置、电子设备及存储介质 Download PDFInfo
- Publication number
- WO2019148923A1 WO2019148923A1 PCT/CN2018/114392 CN2018114392W WO2019148923A1 WO 2019148923 A1 WO2019148923 A1 WO 2019148923A1 CN 2018114392 W CN2018114392 W CN 2018114392W WO 2019148923 A1 WO2019148923 A1 WO 2019148923A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- target object
- detected
- target
- preset
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000012545 processing Methods 0.000 claims description 33
- 238000004891 communication Methods 0.000 claims description 19
- 238000001514 detection method Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 11
- 238000012544 monitoring process Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011982 device technology Methods 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000013341 scale-up Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/532—Query formulation, e.g. graphical querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
Definitions
- the present application relates to the field of image processing technologies, and in particular, to a map search method, device, electronic device, and storage medium.
- the location of the target object may be determined in the original image to be detected, and by means of frame selection, etc.
- the location of the target object is indicated in the image to be detected and displayed to the user.
- the user may query or search the target image in the target frame to obtain information about the target object.
- a plurality of target objects are often present in an original image to be detected, and when the positions of the plurality of target objects are relatively close, the position of the target object is determined by the frame selection method, which may cause the target frame. Interference with each other is not conducive to the user to view and select the target object of interest or pending.
- the image in the target frame may contain partial images of other target objects, so that accurate search results cannot be obtained.
- An object of the present application is to provide a map search method, a device, an electronic device, and a storage medium, so as to realize that the detected images of the plurality of target objects are independent of the original image to be detected, and the display is performed separately.
- the user can view and select the target object that is interested or to be processed more intuitively and conveniently, and can improve the search because the image of the plurality of target objects can be independent of the original image to be detected, thereby searching through the image of the target object. accuracy.
- the embodiment of the present application provides a method for searching for a picture, including:
- Searching is performed in the preset database to determine an image that matches the image of the target object to be searched.
- the detecting by using a preset algorithm, the multiple target objects in the image to be detected, including:
- the target object detection network trained by the depth learning based method detects the image to be detected and detects the plurality of target objects in the image to be detected.
- the method before the step of displaying the multiple target object images in the preset position, the method further includes:
- the displaying the plurality of target object images respectively at the preset position comprises:
- a plurality of target object images subjected to image enhancement processing are respectively displayed at preset positions.
- the method before the step of displaying the multiple target object images in the preset position, the method further includes:
- the displaying the plurality of target object images respectively at the preset position comprises:
- a plurality of target object images subjected to the scaling process are respectively displayed at preset positions.
- the determining, by using a preset algorithm, the multiple target objects in the to-be-detected image, and determining coordinate information of an area where the multiple target objects respectively belong including:
- the searching in the preset database determines an image that matches the image of the target object to be searched, including:
- the displaying the plurality of target object images respectively at the preset position including:
- a plurality of target object images corresponding to different to-be-detected images are respectively displayed at a plurality of preset positions.
- the displaying the multiple target object images in the preset position respectively includes:
- the embodiment of the present application further provides a map searching device, including:
- An acquiring module configured to obtain an image to be detected, where the image to be detected includes multiple target objects;
- a detecting module configured to detect the plurality of target objects in the image to be detected by using a preset algorithm, and determine coordinate information of an area where the plurality of target objects respectively exist;
- An extracting module configured to extract, according to the coordinate information, a plurality of target object images respectively corresponding to the plurality of target objects from the image to be detected;
- a display module configured to respectively display the plurality of target object images in a preset position
- a selection module configured to determine, in the plurality of target object images displayed by the preset position, an image of the target object to be searched;
- a searching module configured to perform a search in the preset database to determine an image that matches the image of the target object to be searched.
- the detecting module is specifically configured to:
- the target object detection network trained by the depth learning based method detects the image to be detected and detects the plurality of target objects in the image to be detected.
- the device further includes:
- An image enhancement module configured to separately perform image enhancement processing on the plurality of target object images
- the display module is specifically configured to:
- a plurality of target object images subjected to image enhancement processing are respectively displayed at preset positions.
- the device further includes:
- An image scaling module configured to separately perform scaling processing on the plurality of target object images
- the display module is specifically configured to:
- a plurality of target object images subjected to the scaling process are respectively displayed at preset positions.
- the detecting module is specifically configured to:
- search module is specifically configured to:
- a feature point of the target object corresponding to the image to be searched is used as a feature point of the image of the target image to be searched; and a target corresponding to the image of the target object to be searched is established by using feature points of the image of the target object to be searched
- the model is compared with the image object model in the preset database to determine an image that matches the image of the target object to be searched.
- the display module is specifically configured to:
- a plurality of target object images corresponding to different to-be-detected images are respectively displayed at a plurality of preset positions.
- the display module is specifically configured to:
- the embodiment of the present application further provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus;
- a memory for storing a computer program
- the processor when used to execute a program stored on the memory, implements a map search method of any of the above.
- a computer readable storage medium stores instructions that, when run on a computer, cause the computer to perform any of the Figure search method.
- the embodiment of the present application further provides a computer program product comprising instructions, when executed on a computer, causing the computer to perform the graph search method of any of the above.
- the image searching method, the device, the electronic device and the storage medium provided by the embodiment of the present invention can detect a plurality of target objects from the image to be detected through a preset algorithm after obtaining the image to be detected, and determine each The coordinate information of the area where a target object is located.
- the pixel information of the region where the target object is located is extracted from each of the target objects from the image to be detected, and the target object image corresponding to the target object is created, and then the plurality of target object images are respectively displayed at the preset position. It is possible to avoid marking a plurality of target objects by means of frame selection or the like in the image to be detected, thereby avoiding a situation in which the target frames overlap each other when the plurality of target objects in the image to be detected are in close proximity.
- the user can view the detected target object more clearly and intuitively, and facilitate the user to perform the image of the plurality of target objects displayed. select.
- the respective target object images are extracted from the image to be detected and are independent of each other, when the search is performed based on the target object image, it is possible to obtain a more accurate search result without interference from other target objects.
- implementing any of the products or methods of the present application necessarily does not necessarily require all of the advantages described above to be achieved at the same time.
- FIG. 1 is a flowchart of a method for searching for a map provided by an embodiment of the present application
- FIG. 2 is another flowchart of a method for searching for a map provided by an embodiment of the present application
- FIG. 3 is a schematic diagram of feature points provided by an embodiment of the present application.
- FIG. 4 is a structural diagram of a map searching device according to an embodiment of the present application.
- FIG. 5 is a structural diagram of an electronic device according to an embodiment of the present application.
- the image search method provided by the embodiment of the present application may separately generate a target object image for each target object after detecting a plurality of target objects in the image to be detected. And displaying a plurality of target object images corresponding to the plurality of target objects respectively in a preset position that is preset and convenient for the user to view, thereby facilitating the user to more intuitively view the detected target object, and is convenient for the user to Among the plurality of target object images displayed, the target object image of interest is determined, and a search is performed in the preset database to search for an image matching the target object image.
- the following is a detailed introduction of the map search method provided by the embodiment of the present application.
- FIG. 1 is a flowchart of a method for searching a map according to an embodiment of the present application, including:
- Step 110 Obtain an image to be detected, where the image to be detected includes a plurality of target objects.
- the embodiments of the present application can be used for various electronic devices having image information processing functions, such as a mobile phone, a computer, a server, a surveillance camera, and a monitoring system to which a surveillance camera is connected.
- the image to be detected may be various images or pictures containing one or more target objects, which may be various types of targets set in advance, for example, may be face targets, vehicles, license plates, or other types of targets.
- the image to be detected may be a monitoring image taken by the surveillance camera, which may include a target object in the monitoring scene, such as a face target or a vehicle, etc.; or the image to be detected may also be a photo, and the photo may contain one or Multiple target objects; or the image to be detected may also be a screenshot screen of a video or the like.
- the electronic device can obtain the image to be detected in various ways, for example, manually input the image to be detected, or the electronic device can take a video or a photo in real time as the image to be detected through the image acquiring device that it has.
- the embodiment of the present application is applied to the monitoring field, the monitoring image captured by the camera can be directly used as the image to be detected, and the image searching method provided by the embodiment of the present application is executed in the camera or the monitoring system to which the camera is connected.
- Step 120 The preset algorithm detects a plurality of target objects in the image to be detected, and determines coordinate information of an area where the plurality of target objects respectively exist.
- the preset algorithm may be a neural network algorithm or other detection algorithms, for example, a principal component analysis algorithm, an independent component analysis algorithm, a singular value feature based algorithm, and the like.
- the electronic device can detect the detected image by using the foregoing preset algorithm, thereby detecting a plurality of target objects in the image to be detected. After detecting a plurality of target objects, the area in which each target object is located in the image to be detected may be separately determined.
- the determined target object in the image to be detected may be an area having a fixed shape including the target object.
- an area having a fixed shape may be a rectangular frame, the rectangular frame includes the target object, and the size of the rectangular frame is adapted to the area of the target object, and all pixel points of the target object are included in the rectangular frame.
- a boundary between the target object and the background image in the image to be detected may be further determined.
- the area where the target object is located in the image to be detected may also be an area containing only the target object, and the area is an irregular area, that is, the boundary of the area is the boundary between the target object and the background image. Thus, in this area, there may be no background image other than the target object.
- coordinate information of the area can be obtained for the area where each target object is located.
- the coordinate information may be an array of area coordinates composed of the coordinates of all the pixels in the area, and may also be a coordinate range of all the pixels in the area.
- the region is a region having a fixed shape, for example, a rectangular region
- the region range of the rectangle can be defined by the coordinates of the four vertices of the rectangle. Therefore, the coordinates of the four vertices can be used to identify the coordinate range of the region, so that the coordinate information can only contain the coordinates of the four vertices, thereby simplifying the coordinate information.
- Step 130 Extract, according to the coordinate information, each pixel point of the region where the plurality of target objects are located from the image to be detected, and form a plurality of target object images respectively corresponding to the plurality of target objects.
- each pixel point of the region where each target object is located may be extracted according to the coordinate information.
- the process of extracting pixels may be to copy all the pixels in the area and copy the attribute information possessed by the pixels.
- the target object image is a sub-image extracted from the image to be detected and independent of the image to be detected.
- mapping methods may also be used to extract and construct a plurality of target object images from the image to be detected.
- the method can be applied to the method for searching in the image provided by the embodiment of the present application, and all of them belong to the protection scope of the embodiment of the present application.
- step 140 a plurality of target object images are respectively displayed at preset positions.
- the respective target object images may be respectively displayed on the preset positions convenient for the user to view.
- the preset position may be other than the image to be detected, for example, may be in the display device, independent of other areas than the image to be detected, such as other windows or frames in the display screen other than displaying the image to be detected.
- the preset position may also be an area selected in advance in the image to be detected, for example, a lowermost area of the image to be detected, or the like.
- a plurality of target object images When a plurality of target object images are displayed, a plurality of target object images may be sequentially displayed in a preset position, thereby enabling the user to visually and clearly view the extracted respective target object images.
- Step 150 Determine, in the plurality of target object images displayed by the preset position, an image of the target object to be searched.
- the electronic device can display a plurality of target object images to the user at preset positions.
- the user can select an image of the target object to be searched for.
- a plurality of target objects can be detected in the monitoring screen, and the target object can be a face target.
- the relevant person may select a target object image of the suspicious person from the plurality of target object images displayed by the preset position, and use the target object image as the target object image to be searched, and search for the image of the target object to be searched by the subsequent steps.
- Identity information may be used to select a target object image of the suspicious person from the plurality of target object images displayed by the preset position, and use the target object image as the target object image to be searched, and search for the image of the target object to be searched by the subsequent steps.
- the electronic device can determine the target object image to be searched from among the plurality of target object images displayed.
- the user's selection may be such that the electronic device receives the user's selection signal by means of a click operation, etc., thereby determining one or more target object images selected by the user, and using the one or more target object images as the to-be-searched. Target object image.
- Step 160 Perform a search in the preset database to determine an image that matches the image of the target object to be searched.
- the preset database may be a database in which a large amount of image information is stored, including image data, and corresponding information.
- a large amount of personnel information can be saved in the preset database.
- the personnel information saved in the preset database includes at least a person's avatar and identity information.
- the avatar can be photographed and registered, and the captured photos and registered identity information can be entered into the database, and the database can be used as a preset database corresponding to the personnel in the specific place.
- information of various vehicles may be saved in the preset data, including a picture, a name, a model, a configuration, and the like of the vehicle.
- a search can be performed in the preset database.
- an image matching the image of the target object to be searched can be obtained.
- the information corresponding to the image in the preset database can be used as the information of the image of the target object to be searched, thereby realizing the function of searching by image.
- the target object is a face target
- the related data of the avatar saved in the preset database may be matched.
- the face target can be considered to match the avatar, that is, the face target
- the corresponding target object image is the same person as the avatar. Therefore, the identity information corresponding to the avatar matching the face target can be used as the identity information of the face target image.
- the matching image may be searched in the preset database by the above manner, and information about the image may be obtained.
- a plurality of target objects are detected from the image to be detected by using a preset algorithm, and coordinate information of an area where each target object is located is determined. Through the coordinate information, pixel points of the region where the target object is located are extracted from the image to be detected for each target object, and a target object image corresponding to the target object is created, and then a plurality of target object images are respectively displayed at the preset position. It is possible to avoid marking a plurality of target objects by means of frame selection or the like in the image to be detected, thereby avoiding a situation in which the target frames overlap each other when the plurality of target objects in the image to be detected are in close proximity.
- the user can view the detected target object more clearly and intuitively, and facilitate the user to perform the image of the plurality of target objects displayed. select. Moreover, since the respective target object images are extracted from the image to be detected and are independent of each other, when the search is performed based on the target object image, it is possible to obtain a more accurate search result without interference from other target objects.
- step 120 detecting a plurality of target objects in the image to be detected by using a preset algorithm may include:
- the target object detection network trained by the deep learning-based method detects the detected image and detects a plurality of target objects in the image to be detected.
- the initial convolutional neural network can be trained through a large number of samples containing the target object, which can make the convolutional neural network have the ability to detect the target object.
- the convolutional neural network can then be used as the target object detection network.
- the target object detection network may be stored in the electronic device or stored in a server or a service cloud platform corresponding to the electronic device.
- the electronic device can directly run the target object detection network to detect the acquired image to be detected, or perform interaction with the data of the server or the service cloud platform to complete the detection of the image to be detected.
- the electronic device may input the image to be detected into the target object detection network, and the target object detection network may detect and calculate the image to be detected, and use the target object to detect the feature of the target object that is trained within the network, All the features included in the image to be detected are matched, and then a plurality of target objects included in the image to be detected are detected.
- the target object detection network can detect multiple target objects included in the image to be detected more quickly and accurately. Thereby, the accuracy of the extracted target object image can be further improved, and the overall operational efficiency is improved.
- the method may further include:
- Step 170 Perform image enhancement processing on each of the plurality of target object images.
- the electronic device constructs and obtains a plurality of target object images
- the plurality of target object images are obtained by copying the pixel points
- the degree of clarity depends only on the sharpness of the image to be detected. And often the target object occupies a small area in the image to be detected, so that the obtained plurality of target object images are not sharp in definition.
- image enhancement is a technical means of making images that are not clearly visible clear.
- Image quality, rich information, and image interpretation and detection can be enhanced by means of frequency domain method or spatial domain method.
- step 140 multiple target object images are respectively displayed at the preset position, including:
- a plurality of target object images subjected to image enhancement processing are respectively displayed at preset positions.
- the plurality of target object images subjected to the image enhancement processing are displayed at the preset position, thereby improving the display effect of the target object image.
- the method may further include:
- step 180 the plurality of target object images are respectively subjected to scaling processing.
- the obtained plurality of target object images may be image-scaled. Therefore, the image of the target object with a smaller original area can be adjusted to a larger target image of the target object that is more convenient for the user to view. Or, the original larger target object image is reduced to a smaller target object image that is convenient for the user to view.
- the target object image subjected to the image enhancement processing may be subjected to scaling processing, thereby further improving the image quality of the target object image.
- step 140 multiple target object images are respectively displayed at the preset position, including:
- a plurality of target object images subjected to the scaling process are respectively displayed at preset positions.
- the plurality of target object images subjected to the scaling process may be displayed at a preset position to improve the display effect of the target object image.
- the image quality of the target object image may be improved, and the processed plurality of target object images are performed at preset positions.
- step 120 a plurality of target objects are detected in the image to be detected by using a preset algorithm, and multiple targets are determined.
- the coordinate information of the area where the objects are located including:
- a plurality of target objects are detected in the image to be detected, and coordinate information of regions in which the plurality of target objects are located and feature points respectively possessed by the plurality of target objects are determined.
- a preset algorithm may be used to determine feature points respectively of the plurality of target objects.
- the feature points that the target object has are the key points that can directly reflect the characteristics of the target object.
- the dot in FIG. 3 is a feature point in the face.
- the feature point is mainly a pixel point of a key position of the face, and the feature point can reflect the positional relationship between the main features of the face, and the proportional relationship and other parameters.
- the target object is another object, such as a vehicle, the contour point of the vehicle, the position of the lamp position, and the like may be used as feature points of the target object.
- step 160 performing a search in the preset database to determine an image that matches the image of the target object to be searched may include:
- Step 161 The feature point of the target object corresponding to the image of the target object to be searched is used as the feature point of the image of the target object to be searched.
- the image features in the target object image are relatively The image feature in the target object does not change, so the feature point of the target object corresponding to the image of the target object to be searched can be directly used as the feature point of the image of the target object to be searched.
- the corresponding feature points may also perform corresponding image enhancement processing or scale up or down the relative position distance between the feature points.
- Step 162 The target model corresponding to the image of the target object to be searched is established by using the feature points of the image of the target object to be searched, and compared with the image object model in the preset database, and the image of the target object to be searched is determined to be matched. Image.
- the target model is a mathematical model constructed for the target object image by using feature points in the image of the target object to be searched.
- the feature points can reflect the relative position and proportional relationship between the facial features and the like. Image matching search can be performed more conveniently and quickly through the target model.
- the target model can be used to search in the default database.
- the default database can hold an object model with a large number of target objects.
- the target object is a face target
- a large number of face targets and corresponding information may be saved in the preset database in the form of a face target model. Therefore, the established target model can be directly used to match the object model in the preset database.
- an image matching the image of the target object to be searched is quickly determined. Further, related information corresponding to the image is obtained.
- the positional relationship between the main features of the target object reflected by the feature points, and the proportional relationship and other parameters can be utilized to quickly establish a target model corresponding to the image of the target object to be searched. Therefore, the efficiency of the search can be improved while improving the efficiency of the entire search process.
- step 140 displaying a plurality of target object images at preset positions may include:
- a plurality of target object images corresponding to different images to be detected are respectively displayed at a plurality of preset positions.
- the electronic device may acquire a plurality of to-be-detected images in parallel, and may simultaneously process the plurality of to-be-detected images according to the image search method provided by the embodiment of the present application, or serially and sequentially press the plurality of to-be-detected images.
- the processing provided by the application embodiment is separately processed by a map search method.
- a target object image corresponding to a plurality of target objects included in the image to be detected may be displayed at a preset position corresponding to the image to be detected. That is, there may be a plurality of preset positions, and each preset position corresponds to one image to be detected. Thereby, the user can more intuitively view a plurality of target object images corresponding to different images to be detected.
- step 140 respectively displaying the multiple target object images in a preset position, may include:
- a plurality of target object images within a preset number are respectively displayed at preset positions, and the preset number is a maximum number of target objects that can be detected by the preset algorithm at one time.
- the number of target objects contained in the image to be detected can be clearly displayed.
- the preset number of target object images that can be displayed by the preset position can enable the user to clearly understand the maximum number of target objects that the preset algorithm can detect at one time, and clearly display the preset algorithm. The performance of the operation.
- the target object image determined later may be used to cover the target object image that has been previously displayed. Thereby, the number of target object images at the preset position does not exceed a preset number. If the corresponding target object images in the respective to-be-detected images are respectively displayed at different preset positions, the target object images displayed by each preset position do not exceed the preset number.
- FIG. 4 is a structural diagram of a map searching device according to an embodiment of the present application, including:
- An obtaining module 401 configured to obtain an image to be detected, where the image to be detected includes multiple target objects;
- the detecting module 402 is configured to detect the plurality of target objects in the image to be detected by using a preset algorithm, and determine coordinate information of an area where the plurality of target objects respectively exist;
- the extracting module 403 is configured to respectively extract, from the image to be detected, a plurality of target object images respectively corresponding to the plurality of target objects according to the coordinate information;
- a display module 404 configured to respectively display the plurality of target object images in a preset position
- a selection module 405, configured to determine, in the plurality of target object images displayed by the preset position, an image of the target object to be searched;
- the searching module 406 is configured to perform a search in the preset database to determine an image that matches the image of the target object to be searched.
- a plurality of target objects are detected from the image to be detected by using a preset algorithm, and coordinate information of an area where each target object is located is determined. Through the coordinate information, pixel points of the region where the target object is located are extracted from the image to be detected for each target object, and a target object image corresponding to the target object is created, and then a plurality of target object images are respectively displayed at the preset position. It is possible to avoid marking a plurality of target objects by means of frame selection or the like in the image to be detected, thereby avoiding a situation in which the target frames overlap each other when the plurality of target objects in the image to be detected are in close proximity.
- the user can view the detected target object more clearly and intuitively, and facilitate the user to perform the image of the plurality of target objects displayed. select. Moreover, since the respective target object images are extracted from the image to be detected and are independent of each other, when the search is performed based on the target object image, it is possible to obtain a more accurate search result without interference from other target objects.
- the detecting module 402 is specifically configured to:
- the target object detection network trained by the depth learning based method detects the image to be detected and detects the plurality of target objects in the image to be detected.
- the device further includes:
- An image enhancement module configured to separately perform image enhancement processing on the plurality of target object images
- the display module 404 is specifically configured to:
- a plurality of target object images subjected to image enhancement processing are respectively displayed at preset positions.
- the device further includes:
- An image scaling module configured to separately perform scaling processing on the plurality of target object images
- the display module 404 is specifically configured to:
- a plurality of target object images subjected to the scaling process are respectively displayed at preset positions.
- the detecting module 402 is specifically configured to:
- searching module 406 is specifically configured to:
- a feature point of the target object corresponding to the image to be searched is used as a feature point of the image of the target image to be searched; and a target corresponding to the image of the target object to be searched is established by using feature points of the image of the target object to be searched
- the model is compared with the image object model in the preset database to determine an image that matches the image of the target object to be searched.
- the display module 404 is specifically configured to:
- a plurality of target object images corresponding to different to-be-detected images are respectively displayed at a plurality of preset positions.
- the display module 404 is specifically configured to:
- the embodiment of the present application further provides an electronic device, as shown in FIG. 5, including a processor 501, a communication interface 502, a memory 503, and a communication bus 504, wherein the processor 501, the communication interface 502, and the memory 503 pass through the communication bus 504.
- a processor 501 a communication interface 502
- a memory 503 a communication bus 504
- the processor 501, the communication interface 502, and the memory 503 pass through the communication bus 504.
- the processor 501 is configured to perform the following steps when executing the program stored on the memory 503:
- Searching is performed in the preset database to determine an image that matches the image of the target object to be searched.
- the communication bus mentioned in the above electronic device may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus.
- the communication bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in the figure, but it does not mean that there is only one bus or one type of bus.
- the communication interface is used for communication between the above electronic device and other devices.
- the memory may include a RAM (Random Access Memory), and may also include NVM (Non-Volatile Memory), such as at least one disk storage.
- the memory may also be at least one storage device located away from the aforementioned processor.
- the above-mentioned processor may be a general-purpose processor, including a CPU (Central Processing Unit), an NP (Network Processor), etc., or may be a DSP (Digital Signal Processor) or an ASIC ( Application Specific Integrated Circuit, FPGA (Field-Programmable Gate Array) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component.
- a CPU Central Processing Unit
- NP Network Processor
- DSP Digital Signal Processor
- ASIC Application Specific Integrated Circuit
- FPGA Field-Programmable Gate Array
- a computer readable storage medium having stored therein instructions that, when run on a computer, cause the computer to perform any of the above embodiments The method of searching by graph.
- the description is relatively simple, and the related method is referred to the method embodiment. Part of the description can be.
- the computer program product includes one or more computer instructions.
- the computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable device.
- the computer instructions can be stored in a computer readable storage medium or transferred from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be from a website site, computer, server or data center Transfer to another website site, computer, server, or data center by wire (eg, coaxial cable, fiber optic, digital subscriber line (DSL), or wireless (eg, infrared, wireless, microwave, etc.).
- the computer readable storage medium can be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that includes one or more available media.
- the usable medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (such as a solid state disk (SSD)).
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
- Processing Or Creating Images (AREA)
Abstract
一种以图搜图方法、装置、电子设备及存储介质，所述方法包括：获得待检测图像（110）；通过预设算法，在待检测图像中检测出多个目标对象，并确定出多个目标对象分别所在的区域的坐标信息（120）；根据坐标信息，从待检测图像中分别提取多个目标对象所在区域的各像素点，并构成与所述多个目标对象分别对应的多个目标对象图像（130）；然后在预设位置分别展示多个目标对象图像（140）；在预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像（150）；在预设数据库中进行搜索，确定出与待搜索目标对象图像匹配的图像（160）。通过提取目标对象图像并独立进行展示，使得根据该目标对象图像进行搜索时，能够不受其他目标对象的干扰，得到更准确的搜索结果。
Description
本申请要求于2018年02月02日提交中国专利局、申请号为201810104827.5申请名称为“一种以图搜图方法、装置、电子设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
本申请涉及图像处理技术领域，特别是涉及一种以图搜图方法、装置、电子设备及存储介质。
随着互联网及智能硬件设备技术的不断发展，在各类交互与应用中，对于图像的处理显得越来越重要。其中，对于图像中的各类目标对象，例如人脸、车牌或其他目标等，进行识别并获得与之相关的信息或数据，一直是重点发展的技术方向。
现有技术中，针对待检测图像中的各类目标对象，进行检测并获取有关的信息或数据时，可以在原始的待检测图像中确定出目标对象所在的位置，并通过框选等方式，在该待检测图像中标示出目标对象所在的位置并向用户进行展示。用户在该待检测图像中选定感兴趣或待处理的目标对象对应的目标框后，可以对该目标框内的目标图像进行查询或搜索，得到与该目标对象有关的信息。
然而，在实际应用中，一张原始的待检测图像中往往存在多个目标对象，并且当多个目标对象的位置相对紧密时，通过框选的方式确定出目标对象的位置，会造成目标框相互重叠干涉，不利于用户查看以及选择感兴趣或待处理的目标对象。并且，当直接对目标框中的目标图像进行查询或搜索时，由于目标框相互重叠干涉，所以该目标框内的图像可能会包含有其他目标对象的部分图像，从而不能获得准确的搜索结果。
发明内容
本申请实施例的目的在于提供一种以图搜图方法、装置、电子设备及存 储介质，以实现将检测出的多个目标对象的图像独立于原始的待检测图片，单独的进行展示，使用户可以更直观方便的查看以及选择感兴趣或待处理的目标对象，并且由于多个目标对象的图像能够独立于原始的待检测图片，从而通过该目标对象的图像进行搜索时，能够提高搜索的准确性。具体技术方案如下：
本申请实施例提供了一种以图搜图方法，包括：
获得待检测图像，所述待检测图像中包括多个目标对象；
通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；
根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；
在预设位置分别展示所述多个目标对象图像；
在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；
在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
可选的，所述通过预设算法，在所述待检测图像中检测出所述多个目标对象，包括：
通过基于深度学习的方法训练得到的目标对象检测网络，对所述待检测图像进行检测，检测出所述待检测图像中的所述多个目标对象。
可选的，所述在预设位置分别展示所述多个目标对象图像的步骤之前，所述方法还包括：
将所述多个目标对象图像分别进行图像增强处理；
相应的，所述在预设位置分别展示所述多个目标对象图像，包括：
在预设位置分别展示经过图像增强处理后的多个目标对象图像。
可选的，所述在预设位置分别展示所述多个目标对象图像的步骤之前， 所述方法还包括：
将所述多个目标对象图像分别进行缩放处理；
相应的，所述在预设位置分别展示所述多个目标对象图像，包括：
在预设位置分别展示经过缩放处理后的多个目标对象图像。
可选的，所述通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息，包括：
通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息和所述多个目标对象分别具有的特征点；
相应的，所述在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像，包括：
将所述待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目标对象图像的特征点；
通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
可选的，当获得多个待检测图像时，所述在预设位置分别展示所述多个目标对象图像，包括：
针对所述多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
可选的，所述在预设位置分别展示所述多个目标对象图像，包括：
在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
本申请实施例还提供了一种以图搜图装置，包括：
获取模块，用于获得待检测图像，所述待检测图像中包括多个目标对象；
检测模块，用于通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；
提取模块，用于根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；
展示模块，用于在预设位置分别展示所述多个目标对象图像；
选择模块，用于在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；
搜索模块，用于在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
可选的，所述检测模块，具体用于：
通过基于深度学习的方法训练得到的目标对象检测网络，对所述待检测图像进行检测，检测出所述待检测图像中的所述多个目标对象。
可选的，所述装置还包括：
图像增强模块，用于将所述多个目标对象图像分别进行图像增强处理；
相应的，所述展示模块，具体用于：
在预设位置分别展示经过图像增强处理后的多个目标对象图像。
可选的，所述装置还包括：
图像缩放模块，用于将所述多个目标对象图像分别进行缩放处理；
相应的，所述展示模块，具体用于：
在预设位置分别展示经过缩放处理后的多个目标对象图像。
可选的，所述检测模块，具体用于：
通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息和所述多个目标对象分别具有的特征点；
相应的，所述搜索模块，具体用于：
将所述待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目标对象图像的特征点；通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
可选的，当获得多个待检测图像时，所述展示模块，具体用于：
针对所述多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
可选的，所述展示模块，具体用于：
在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
本申请实施例还提供了一种电子设备，包括处理器、通信接口、存储器和通信总线，其中，处理器，通信接口，存储器通过通信总线完成相互间的通信；
存储器，用于存放计算机程序；
处理器，用于执行存储器上所存放的程序时，实现上述任一种的以图搜图方法。
在本申请实施的又一方面，还提供了一种计算机可读存储介质，所述计算机可读存储介质中存储有指令，当其在计算机上运行时，使得计算机执行上述任一所述的以图搜图方法。
在本申请实施的又一方面，本申请实施例还提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行上述任一所述的以图搜图方法。
本申请实施例提供的一种以图搜图方法、装置、电子设备及存储介质，可以在获得待检测图像之后，通过预设算法从待检测图像中检测出多个目标对象，并且确定出每一个目标对象所在区域的坐标信息。通过该坐标信息，从待检测图像中针对每一个目标对象提取该目标对象所在区域的像素点，并建立该目标对象对应的目标对象图像，然后在预设位置处分别展示多个目标 对象图像。可以避免在待检测图像中通过框选等方式标示出多个目标对象，从而可以避免当待检测图像中的多个目标对象距离很近时，造成的目标框相互重叠干涉等情况。通过将独立于待检测图像的多个目标对象图像分别在预设位置处进行展示，使得用户可以更加清楚直观的查看所检测出的目标对象，并且方便用户对所展示的多个目标对象图像进行选择。并且由于将各个目标对象图像从待检测图像中提取出来，并且相互独立，从而使得根据该目标对象图像进行搜索时，能够不受其他目标对象的干扰，进而得到更加准确的搜索结果。当然，实施本申请的任一产品或方法必不一定需要同时达到以上所述的所有优点。
为了更清楚地说明本申请实施例和现有技术的技术方案，下面对实施例和现有技术中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的以图搜图方法的一种流程图；
图2为本申请实施例提供的以图搜图方法的另一种流程图；
图3为本申请实施例提供的特征点示意图；
图4为本申请实施例提供的以图搜图装置的结构图；
图5为本申请实施例提供的电子设备的结构图。
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请实施例提供的以图搜图方法，可以在检测出待检测图像中多个目 标对象后，针对每一个目标对象单独生成目标对象图像。并且在预设设定好的、方便用户查看的预设位置上展示出多个目标对象分别对应的多个目标对象图像，进而方便用户更直观的查看所检测出的目标对象，并且方便用户从所展示的多个目标对象图像中，确定出感兴趣的目标对象图像，在预设数据库中进行搜索，进而搜索出与该目标对象图像相匹配的图像。下面对本申请实施例提供的以图搜图方法进行具体的介绍。
参见图1，图1为本申请实施例提供的以图搜图方法的一种流程图，包括：
步骤110，获得待检测图像，待检测图像中包括多个目标对象。
本申请实施例可以用于各类具有图像信息处理功能的电子设备，例如，手机、电脑、服务器、监控摄像头以及监控摄像头所连接的监控系统等等。
待检测图像可以为含有一个或多个目标对象的各种图像或图片，目标对象可以是预先设定的各种类型的目标，例如，可以为人脸目标、车辆、车牌、或者其他类型的目标。例如，待检测图像可以为监控摄像头所拍摄的监控画面，其中可以含有监控场景中的目标对象，如人脸目标或车辆等等；或者待检测图像也可以是照片，该照片中可以含有一个或多个目标对象；或者待检测图像还可以是视频的截图画面等等。
电子设备获得待检测图像可以有多种方式，例如，人工输入待检测图像，或者电子设备可以通过自身所具有的图像获取装置，将实时拍摄视频或照片作为待检测图像。当本申请实施例应用于监控领域时，摄像头所拍摄的监控画面可以直接作为待检测图像，并且在摄像头中或者摄像头所连接的监控系统中，执行本申请实施例提供的以图搜图方法。
步骤120，通过预设算法，在待检测图像中检测出多个目标对象，并确定出该多个目标对象分别所在的区域的坐标信息。
预设算法，可以是神经网络算法，或者是其他检测算法，例如，主成分分析算法、独立成分分析算法、基于奇异值特征算法等等。
电子设备利用上述预设算法，可以对待检测图像进行检测，从而在待检测图像中检测出多个目标对象。当检测出多个目标对象之后，可以分别确定出每一个目标对象在待检测图像中所处的区域。
在本申请实施例的一种实施方式中，通过预设算法，确定出的目标对象在待检测图像中所处的区域可以为包括该目标对象的具有固定形状的区域。例如，具有固定形状的区域可以为矩形框，该矩形框中包含有该目标对象，且该矩形框的大小与目标对象的面积相适应，保证目标对象的所有像素点都被包含在该矩形框内。
在本申请实施例的另一种实施方式中，通过预设算法，检测出目标对象后，还可以进一步确定出该目标对象与待检测图像中的背景图像之间的边界。相应的，目标对象在待检测图像中所处的区域还可以为仅含有目标对象的区域，该区域为不规则的区域，即该区域的边界为该目标对象与背景图像之间的边界。从而在该区域中，可以不含有任何除目标对象以外的背景图像。
当确定出每一个目标对象在待检测图像中所处的区域之后，就可以针对每一个目标对象所在的区域，得到该区域的坐标信息。
坐标信息可以为该区域中所有像素点的坐标所组成的区域坐标数组，还可以是该区域中所有像素点的坐标范围。当该区域为具有固定性状的区域时，例如为矩形区域时，则可以用矩形的4个顶点的坐标来划定该矩形的区域范围。所以该4个顶点的坐标可以用于标识出该区域的坐标范围，使得坐标信息可以仅包含该4个顶点的坐标，从而简化坐标信息。
步骤130，根据坐标信息，从待检测图像中分别提取多个目标对象所在区域的各像素点，并构成与多个目标对象分别对应的多个目标对象图像。
电子设备确定出每一个目标对象对应的坐标信息之后，就可以根据该坐标信息，提取每一个目标对象所在区域的各像素点。提取像素点的过程可以是对该区域内所有像素点进行复制，并且复制像素点所具有的属性信息。
通过所提取的像素点，可以构建出每一个目标对象分别对应的目标对象图像。该目标对象图像为从待检测图像中提取的，且独立于待检测图像的子图像。
在实际应用中，还可以采用其他现有的抠图方法，从待检测图像中提取并构建出多个目标对象图像。只要能够满足本申请实施例中的需求，都可以应用于本申请实施例提供的以图搜图方法中，且都属于本申请实施例的保护 范围。
步骤140，在预设位置分别展示多个目标对象图像。
当电子设备已经得到多个目标对象对应的多个目标对象图像之后，就可以在方便用户查看的预设位置上分别展示各个目标对象图像。预设位置可以是待检测图像以外的其他位置，例如可以是在显示装置中，独立于待检测图像之外的其他区域，如，显示屏中除显示待检测图像之外的其他窗口或图框。预设位置也可以是在待检测图像中提前选定的区域，例如，待检测图像的最下侧区域等。
在显示多个目标对象图像时，多个目标对象图像可以在预设位置中按顺序排列展示，从而使用户能够直观且清楚的查看提取出的各个目标对象图像。
步骤150，在预设位置所展示的多个目标对象图像中，确定出待搜索目标对象图像。
电子设备可以在预设位置上向用户展示出多个目标对象图像。用户在对所展示的多个目标对象图像进行查看时，可以从中选择出感兴趣的待搜索目标对象图像。例如，在安防领域中，可以在监控画面中检测出多个目标对象，该目标对象可以为人脸目标。有关人员可以从预设位置所展示的多个目标对象图像中选择出可疑人员的目标对象图像，并将该目标对象图像作为待搜索目标对象图像，通过后续步骤搜索出该待搜索目标对象图像对应的身份信息。
通过用户的选择，电子设备可以从所展示的多个目标对象图像中，确定出待搜索目标对象图像。用户的选择，可以是通过点击操作等方式，使电子设备接收到用户的选择信号，进而确定出用户所选择的一个或多个目标对象图像，并将该一个或多个目标对象图像作为待搜索目标对象图像。
步骤160，在预设数据库中进行搜索，确定出与待搜索目标对象图像匹配的图像。
预设数据库可以是保存有大量的图像信息的数据库，其中包括有图像数据，以及对应的信息。例如，当目标对象为人脸目标时，该预设数据库中可以保存有大量的人员信息。该预设数据库中所保存的人员信息至少包括人员的头像和身份信息。如，在进入特定场所时，可以对访客进行头像拍照以及 身份登记，并将所拍摄的照片及登记的身份信息录入数据库，该数据库就可以作为一个对应该特定场所内人员的预设数据库。或者，例如，当目标对象为车辆时，该预设数据中可以保存有各种车辆的信息，包括车的图片、名称、型号、配置等等。
根据待搜索目标对象图像，可以在预设数据库中进行搜索。通过待搜索目标对象图像，与数据库中的图像数据进行匹配，可以得到与该待搜索目标对象图像相匹配的图像。预设数据库中该图像对应的信息，就可以作为该待搜索目标对象图像的信息，从而实现了以图搜图的功能。
例如，目标对象为人脸目标，当选择出待搜索的人脸目标图像后，可以与预设数据库中所保存的头像的有关数据进行匹配。当匹配成功时，例如，该人脸目标的特征参数与预设数据库中所保存的某个头像的特征参数全部吻合，则可以认为该人脸目标与该头像相匹配，即该人脸目标所对应的目标对象图像与该头像为同一人。所以，与该人脸目标相匹配的头像所对应的身份信息，可以作为该人脸目标图像的身份信息。同样的，当目标对象为其他类型时，也可以通过上述方式，在预设数据库中搜索出相匹配的图像，并获得该图像有关的信息。
在本申请实施例中，在获得待检测图像之后，通过预设算法从待检测图像中检测出多个目标对象，并且确定出每一个目标对象所在区域的坐标信息。通过该坐标信息，从待检测图像中针对每一个目标对象提取该目标对象所在区域的像素点，并建立该目标对象对应的目标对象图像，然后在预设位置处分别展示多个目标对象图像。可以避免在待检测图像中通过框选等方式标示出多个目标对象，从而可以避免当待检测图像中的多个目标对象距离很近时，造成的目标框相互重叠干涉等情况。通过将独立于待检测图像的多个目标对象图像分别在预设位置处进行展示，使得用户可以更加清楚直观的查看所检测出的目标对象，并且方便用户对所展示的多个目标对象图像进行选择。并且由于将各个目标对象图像从待检测图像中提取出来，并且相互独立，从而使得根据该目标对象图像进行搜索时，能够不受其他目标对象的干扰，进而得到更加准确的搜索结果。
结合上面的实施例，步骤120中的，通过预设算法，在待检测图像中检测出多个目标对象，可以包括：
通过基于深度学习的方法训练得到的目标对象检测网络，对待检测图像进行检测，检测出待检测图像中的多个目标对象。
可以通过大量的包含目标对象的样本对初始的卷积神经网络进行训练，可以使得该卷积神经网络具有检测出目标对象的能力。然后就可以将该卷积神经网络作为目标对象检测网络。该目标对象检测网络可以存储于电子设备内部，或者存储于电子设备所对应的服务器或服务云平台。电子设备可以直接运行该目标对象检测网络对所获取的待检测图像进行检测，或者通过与服务器或服务云平台的数据交互，完成对待检测图像的检测。
具体的，电子设备可以将待检测图像输入该目标对象检测网络，目标对象检测网络可以对该待检测图像进行检测及相应的计算，利用目标对象检测网络内部经过训练得到的目标对象的特征，对该待检测图像所含有的全部特征进行匹配，进而检测出该待检测图像中所包括的多个目标对象。
相比于现有的其他图像检测技术，利用目标对象检测网络可以更加快速准确的检测出待检测图像中所包括的多个目标对象。从而可以进一步提高所提取的目标对象图像的准确度，并且提高整体的运行效率。
结合上述的各实施例，参见图2，在本申请实施例的一种实施方式中，在步骤140，在预设位置分别展示多个目标对象图像之前，所述方法还可以包括：
步骤170，将多个目标对象图像分别进行图像增强处理。
电子设备构建并得到多个目标对象图像之后，由于该多个目标对象图像是通过像素点的复制得到的，所以其清晰程度仅取决于待检测图像的清晰度。并且往往目标对象在待检测图像中所占面积较小，所以导致所获得的多个目标对象图像的清晰度不高。
所以，可以对多个目标对象图像分别采用图像增强技术进行处理，从而提高多个目标对象图像的清晰度。图像增强是一种将原来不清晰的图像变得 清晰的技术手段。可以通过频率域法或空间域法等方式，改善图像质量、丰富信息量，加强图像判读和检测效果。
相应的，步骤140，在预设位置分别展示多个目标对象图像，包括：
在预设位置分别展示经过图像增强处理后的多个目标对象图像。
当对多个目标对象图像完成图像增强处理后，再将该经过图像增强处理的多个目标对象图像在预设位置处进行展示，可以提高目标对象图像的展示效果。
在本申请实施例的另一种实施方式中，在步骤140，在预设位置分别展示多个目标对象图像之前，所述方法还可以包括：
步骤180，将多个目标对象图像分别进行缩放处理。
同样的，为了提高目标对象图像的展示效果，还可以将所获得的多个目标对象图像进行图像缩放。从而可以将原来面积较小的目标对象图像，调整为面积较大的，更便于用户查看的目标对象图像。或者，将原来较大的目标对象图像，缩小成用户方便查看的较小的目标对象图像。并且，还可以在对多个目标对象图像完成图像增强处理后，再对经过图像增强处理的目标对象图像进行缩放处理，从而进一步提高目标对象图像的图像质量。
相应的，步骤140，在预设位置分别展示多个目标对象图像，包括：
在预设位置分别展示经过缩放处理后的多个目标对象图像。
当对多个目标对象图像分别完成缩放处理后，可以将该经过缩放处理的多个目标对象图像在预设位置处进行展示，以提高目标对象图像的展示效果。在本申请实施例中，通过图像增强或者缩放的方式，将所获得的多个目标对象图像进行处理，可以提高目标对象图像的图像质量，将处理后的多个目标对象图像在预设位置进行展示时，可以使用户看到更清晰、图像质量更好的目标对象图像，进而提高用户体验。
并且，结合前面实施例，在预设位置展示多个目标对象图像之前，首先对多个目标对象图像进行图像增强或放大处理，从而使得所展示的多个目标对象图像，较在原来的待检测图像中，其图像质量等有了很大提高，从而在 后续的步骤中，对该目标对象图像进行搜索时，能够更快捷准确的得到搜索结果，进一步提高了搜索的效率和准确性。
结合上述的实施例，可选的，在本申请实施例提供的以图搜图方法中，步骤120，通过预设算法，在待检测图像中检测出多个目标对象，并确定出多个目标对象分别所在的区域的坐标信息，包括：
通过预设算法，在待检测图像中检测出多个目标对象，并确定出多个目标对象分别所在的区域的坐标信息和多个目标对象分别具有的特征点。
在通过预设算法，对待检测图像进行检测，并检测出多个目标对象时，还可以利用预设算法确定出多个目标对象分别具有的特征点。目标对象具有的特征点，是指能够直接反映目标对象的特征的关键点。例如，当目标对象为人脸目标时，参见图3，图3中的圆点为人脸中的特征点。从图3中可以看出，该特征点主要为人脸的关键位置的像素点，通过该特征点可以反映出面部主要特征之间的位置关系，以及比例关系等参数。
或者，目标对象为其他对象，如为车辆时，则该车辆的轮廓点，车灯位置点等等都可以作为该目标对象的特征点。
相应的，在本申请实施例提供的以图搜图方法中，步骤160，在预设数据库中进行搜索，确定出与待搜索目标对象图像匹配的图像，可以包括：
步骤161，将待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目标对象图像的特征点。
确定出每一个目标对象的特征点后，当确定出用户所选择的待搜索目标对象图像时，由于目标对象图像是根据待检测图像中的目标对象直接得到的，目标对象图像中的图像特征相对于目标对象中的图像特征没有发生变化，所以可以直接将该待搜索目标对象图像对应的目标对象的特征点作为该待搜索目标对象图像的特征点。当然，如果该目标对象图像为经过图像增强处理或缩放处理后的目标对象图像，则对应的特征点也可以进行相应的图像增强处理或按比例放大或缩小特征点之间的相对位置距离。
步骤162，通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
目标模型是利用待搜索目标对象图像中的特征点，针对该目标对象图像所构建的数学模型。例如，当目标对象为人脸目标时，特征点可以反映出人脸五官之间的相对位置及比例关系等等。通过该目标模型可以更方便快捷的进行图像匹配搜索。
得到目标模型之后，就可以利用该目标模型，在预设数据库中进行搜索。预设数据库可以保存有大量的目标对象的对象模型。例如，当目标对象为人脸目标时，预设数据库中可以以人脸目标模型的形式保存有大量的人脸目标及相应的信息。从而可以直接采用所建立的目标模型，与预设数据库中的对象模型进行匹配。通过目标模型与对象模型的匹配，从而快速的确定出与待搜索目标对象图像匹配的图像。进而得到该图像对应的有关信息。
通过待搜索目标对象图像的特征点，可以利用特征点所反映出的目标对象主要特征之间的位置关系，以及比例关系等参数，快速的建立待搜索目标对象图像对应的目标模型。从而可以在提高整个搜索过程效率的同时，还可以提高搜索的准确度。
在实际应用中，在很多情况下，需要对多个待检测图像一起进行处理，所以，当获得多个待检测图像时，步骤140，在预设位置分别展示多个目标对象图像，可以包括：
针对多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
电子设备可以获取多个待检测图像，并且可以同时并行对多个待检测图像按本申请实施例提供的以图搜图方法进行处理，也可以按顺序串行的对多个待检测图像按本申请实施例提供的以图搜图方法分别进行处理。
当电子设备完成对多个待检测图像的处理后，为了能够更加便于用户查看根据各个待检测图像中所确定出的目标对象图像。可以针对每一个待检测图像，在该待检测图像对应的预设位置展示该待检测图像中所含有的多个目 标对象对应的目标对象图像。即可以存在多个预设位置，每一个预设位置对应一个待检测图像。从而使得用户能够更加直观的查看不同待检测图像所对应的多个目标对象图像。
可选的，结合上面的各个实施例，步骤140，在预设位置分别展示所述多个目标对象图像，可以包括：
在预设位置分别展示预设数量以内的多个目标对象图像，预设数量为预设算法一次能够检测出的最大数量的目标对象个数。
通过在预设位置分别展示不大于预设数量的多个目标对象图像，可以清楚的显示出，待检测图像中所含有的目标对象的数量。并且，通过预设位置所能够展示的预设数量的多个目标对象图像，可以使用户能够清楚的了解预设算法一次能够检测出的最大数量的目标对象个数，清楚的显示出预设算法的运算性能。
当同时对多个待检测图像进行处理，如果在同一个预设位置展示所有待检测图像中对应的目标对象图像时，可以使用后面确定出的目标对象图像覆盖前面已经展示过的目标对象图像，从而使得该预设位置处的目标对象图像的数量不超过预设数量。如果在不同的预设位置分别展示各个待检测图像中对应的目标对象图像时，每一个预设位置所展示的目标对象图像均不超过预设数量。
参见图4，图4为本申请实施例提供的以图搜图装置的结构图，包括：
获取模块401，用于获得待检测图像，所述待检测图像中包括多个目标对象；
检测模块402，用于通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；
提取模块403，用于根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；
展示模块404，用于在预设位置分别展示所述多个目标对象图像；
选择模块405，用于在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；
搜索模块406，用于在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
在本申请实施例中，在获得待检测图像之后，通过预设算法从待检测图像中检测出多个目标对象，并且确定出每一个目标对象所在区域的坐标信息。通过该坐标信息，从待检测图像中针对每一个目标对象提取该目标对象所在区域的像素点，并建立该目标对象对应的目标对象图像，然后在预设位置处分别展示多个目标对象图像。可以避免在待检测图像中通过框选等方式标示出多个目标对象，从而可以避免当待检测图像中的多个目标对象距离很近时，造成的目标框相互重叠干涉等情况。通过将独立于待检测图像的多个目标对象图像分别在预设位置处进行展示，使得用户可以更加清楚直观的查看所检测出的目标对象，并且方便用户对所展示的多个目标对象图像进行选择。并且由于将各个目标对象图像从待检测图像中提取出来，并且相互独立，从而使得根据该目标对象图像进行搜索时，能够不受其他目标对象的干扰，进而得到更加准确的搜索结果。
可选的，在本申请实施例提供的以图搜图装置中，所述检测模块402，具体用于：
通过基于深度学习的方法训练得到的目标对象检测网络，对所述待检测图像进行检测，检测出所述待检测图像中的所述多个目标对象。
可选的，在本申请实施例提供的以图搜图装置中，所述装置还包括：
图像增强模块，用于将所述多个目标对象图像分别进行图像增强处理；
相应的，所述展示模块404，具体用于：
在预设位置分别展示经过图像增强处理后的多个目标对象图像。
可选的，在本申请实施例提供的以图搜图装置中，所述装置还包括：
图像缩放模块，用于将所述多个目标对象图像分别进行缩放处理；
相应的，所述展示模块404，具体用于：
在预设位置分别展示经过缩放处理后的多个目标对象图像。
可选的，在本申请实施例提供的以图搜图装置中，所述检测模块402，具体用于：
通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息和所述多个目标对象分别具有的特征点；
相应的，所述搜索模块406，具体用于：
将所述待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目标对象图像的特征点；通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
可选的，在本申请实施例提供的以图搜图装置中，当获得多个待检测图像时，所述展示模块404，具体用于：
针对所述多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
可选的，在本申请实施例提供的以图搜图装置中，所述展示模块404，具体用于：
在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
本申请实施例还提供了一种电子设备，如图5所示，包括处理器501、通信接口502、存储器503和通信总线504，其中，处理器501，通信接口502，存储器503通过通信总线504完成相互间的通信，
存储器503，用于存放计算机程序；
处理器501，用于执行存储器503上所存放的程序时，实现如下步骤：
获得待检测图像，所述待检测图像中包括多个目标对象；
通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；
根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；
在预设位置分别展示所述多个目标对象图像；
在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；
在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
上述电子设备提到的通信总线可以是PCI(Peripheral Component Interconnect，外设部件互连标准)总线或EISA(Extended Industry Standard Architecture，扩展工业标准结构)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。为便于表示，图中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括RAM(Random Access Memory，随机存取存储器)，也可以包括NVM(Non-Volatile Memory，非易失性存储器)，例如至少一个磁盘存储器。可选的，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器，包括CPU(Central Processing Unit，中央处理器)、NP(Network Processor，网络处理器)等；还可以是DSP(Digital Signal Processor，数字信号处理器)、ASIC(Application Specific Integrated Circuit，专用集成电路)、FPGA(Field-Programmable Gate Array，现场可编程门阵列)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
在本申请提供的又一实施例中，还提供了一种计算机可读存储介质，该 计算机可读存储介质中存储有指令，当其在计算机上运行时，使得计算机执行上述实施例中任一所述的以图搜图方法。
在本申请提供的又一实施例中，还提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行上述实施例中任一所述的以图搜图方法。
对于以图搜图装置、计算机可读存储介质以及计算机程序产品实施例而言，由于其所涉及的方法内容基本相似于前述的方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程设备。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质，(例如，软盘、硬盘、磁带)、光介质(例如，DVD)、或者半导体介质(例如固态硬盘Solid State Disk(SSD))等。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除 在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置等实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等，均包含在本申请的保护范围内。
Claims (16)
- 一种以图搜图方法，其特征在于，包括：获得待检测图像，所述待检测图像中包括多个目标对象；通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；在预设位置分别展示所述多个目标对象图像；在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
- 根据权利要求1所述的方法，其特征在于，所述通过预设算法，在所述待检测图像中检测出所述多个目标对象，包括：通过基于深度学习的方法训练得到的目标对象检测网络，对所述待检测图像进行检测，检测出所述待检测图像中的所述多个目标对象。
- 根据权利要求1所述的方法，其特征在于，所述在预设位置分别展示所述多个目标对象图像的步骤之前，所述方法还包括：将所述多个目标对象图像分别进行图像增强处理；相应的，所述在预设位置分别展示所述多个目标对象图像，包括：在预设位置分别展示经过图像增强处理后的多个目标对象图像。
- 根据权利要求1所述的方法，其特征在于，所述在预设位置分别展示所述多个目标对象图像的步骤之前，所述方法还包括：将所述多个目标对象图像分别进行缩放处理；相应的，所述在预设位置分别展示所述多个目标对象图像，包括：在预设位置分别展示经过缩放处理后的多个目标对象图像。
- 根据权利要求1-4所述的方法，其特征在于，所述通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息，包括：通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息和所述多个目标对象分别具有的特征点；相应的，所述在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像，包括：将所述待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目标对象图像的特征点；通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
- 根据权利要求1所述的方法，其特征在于，当获得多个待检测图像时，所述在预设位置分别展示所述多个目标对象图像，包括：针对所述多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
- 根据权利要求1或6所述的方法，其特征在于，所述在预设位置分别展示所述多个目标对象图像，包括：在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
- 一种以图搜图装置，其特征在于，包括：获取模块，用于获得待检测图像，所述待检测图像中包括多个目标对象；检测模块，用于通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息；提取模块，用于根据所述坐标信息，从所述待检测图像中分别提取所述多个目标对象分别对应的多个目标对象图像；展示模块，用于在预设位置分别展示所述多个目标对象图像；选择模块，用于在所述预设位置所展示的所述多个目标对象图像中，确定出待搜索目标对象图像；搜索模块，用于在预设数据库中进行搜索，确定出与所述待搜索目标对象图像匹配的图像。
- 根据权利要求8所述的装置，其特征在于，所述检测模块，具体用于：通过基于深度学习的方法训练得到的目标对象检测网络，对所述待检测图像进行检测，检测出所述待检测图像中的所述多个目标对象。
- 根据权利要求8所述的装置，其特征在于，所述装置还包括：图像增强模块，用于将所述多个目标对象图像分别进行图像增强处理；相应的，所述展示模块，具体用于：在预设位置分别展示经过图像增强处理后的多个目标对象图像。
- 根据权利要求8所述的装置，其特征在于，所述装置还包括：图像缩放模块，用于将所述多个目标对象图像分别进行缩放处理；相应的，所述展示模块，具体用于：在预设位置分别展示经过缩放处理后的多个目标对象图像。
- 根据权利要求8-11所述的装置，其特征在于，所述检测模块，具体用于：通过预设算法，在所述待检测图像中检测出所述多个目标对象，并确定出所述多个目标对象分别所在的区域的坐标信息和所述多个目标对象分别具有的特征点；相应的，所述搜索模块，具体用于：将所述待搜索目标对象图像对应的目标对象的特征点，作为该待搜索目 标对象图像的特征点；通过所述待搜索目标对象图像的特征点，建立所述待搜索目标对象图像对应的目标模型，与预设数据库中图像对象模型比对，确定出与所述待搜索目标对象图像匹配的图像。
- 根据权利要求8所述的装置，其特征在于，当获得多个待检测图像时，所述展示模块，具体用于：针对所述多个待检测图像，分别在多个预设位置分别展示不同待检测图像所对应的多个目标对象图像。
- 根据权利要求8或13所述的装置，其特征在于，所述展示模块，具体用于：在所述预设位置分别展示预设数量以内的所述多个目标对象图像，所述预设数量为所述预设算法一次能够检测出的最大数量的目标对象个数。
- 一种电子设备，其特征在于，包括处理器、通信接口、存储器和通信总线，其中，处理器，通信接口，存储器通过通信总线完成相互间的通信；存储器，用于存放计算机程序；处理器，用于执行存储器上所存放的程序时，实现权利要求1-7任一所述的方法步骤。
- 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现权利要求1-7任一所述的方法步骤。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810104827.5 | 2018-02-02 | ||
| CN201810104827.5A CN110110110A (zh) | 2018-02-02 | 2018-02-02 | 一种以图搜图方法、装置、电子设备及存储介质 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019148923A1 true WO2019148923A1 (zh) | 2019-08-08 |
Family
ID=67477860
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2018/114392 WO2019148923A1 (zh) | 2018-02-02 | 2018-11-07 | 一种以图搜图方法、装置、电子设备及存储介质 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110110110A (zh) |
| WO (1) | WO2019148923A1 (zh) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111274431A (zh) * | 2020-02-04 | 2020-06-12 | 浙江大华技术股份有限公司 | 一种图像检索处理方法及装置 |
| CN111813979A (zh) * | 2020-07-14 | 2020-10-23 | 杭州海康威视数字技术股份有限公司 | 一种信息检索方法、装置及电子设备 |
| CN112380372A (zh) * | 2020-11-13 | 2021-02-19 | 上海哔哩哔哩科技有限公司 | 搜索图像的方法及计算设备 |
| CN113868453B (zh) * | 2021-09-28 | 2024-02-27 | 北京百度网讯科技有限公司 | 对象推荐方法和装置 |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103491299A (zh) * | 2013-09-17 | 2014-01-01 | 宇龙计算机通信科技(深圳)有限公司 | 一种拍照处理方法和装置 |
| CN106778615A (zh) * | 2016-12-16 | 2017-05-31 | 中新智擎有限公司 | 一种识别用户身份的方法、装置和物业服务机器人 |
| CN106844492A (zh) * | 2016-12-24 | 2017-06-13 | 深圳云天励飞技术有限公司 | 一种人脸识别的方法、客户端、服务器及系统 |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4997175B2 (ja) * | 2008-05-23 | 2012-08-08 | 株式会社Ｐｆｕ | オブジェクト管理装置、携帯端末、および、オブジェクト操作方法 |
| CN103916626A (zh) * | 2013-01-05 | 2014-07-09 | 中兴通讯股份有限公司 | 一种监控录像信息提供方法、装置及视频监控系统 |
| KR20140122952A (ko) * | 2013-04-11 | 2014-10-21 | 삼성전자주식회사 | 이미지 합성 방법 및 이를 구현하는 전자 장치 |
| CN104156689B (zh) * | 2013-05-13 | 2017-03-22 | 浙江大华技术股份有限公司 | 一种对目标对象的特征信息进行定位的方法和设备 |
| CN106372606A (zh) * | 2016-08-31 | 2017-02-01 | 北京旷视科技有限公司 | 目标对象信息生成方法和单元、识别方法和单元及系统 |
-
2018
- 2018-02-02 CN CN201810104827.5A patent/CN110110110A/zh active Pending
- 2018-11-07 WO PCT/CN2018/114392 patent/WO2019148923A1/zh active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103491299A (zh) * | 2013-09-17 | 2014-01-01 | 宇龙计算机通信科技(深圳)有限公司 | 一种拍照处理方法和装置 |
| CN106778615A (zh) * | 2016-12-16 | 2017-05-31 | 中新智擎有限公司 | 一种识别用户身份的方法、装置和物业服务机器人 |
| CN106844492A (zh) * | 2016-12-24 | 2017-06-13 | 深圳云天励飞技术有限公司 | 一种人脸识别的方法、客户端、服务器及系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110110110A (zh) | 2019-08-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376667B (zh) | 目标检测方法、装置及电子设备 | |
| CN109961009B (zh) | 基于深度学习的行人检测方法、系统、装置及存储介质 | |
| WO2019148923A1 (zh) | 一种以图搜图方法、装置、电子设备及存储介质 | |
| US9697416B2 (en) | Object detection using cascaded convolutional neural networks | |
| WO2019218824A1 (zh) | 一种移动轨迹获取方法及其设备、存储介质、终端 | |
| WO2018137623A1 (zh) | 图像处理方法、装置以及电子设备 | |
| CN108734185B (zh) | 图像校验方法和装置 | |
| WO2016187888A1 (zh) | 基于字符识别的关键词通知方法及设备、计算机程序产品 | |
| CN110136198B (zh) | 图像处理方法及其装置、设备和存储介质 | |
| TW202026948A (zh) | 活體檢測方法、裝置以及儲存介質 | |
| WO2018210047A1 (zh) | 数据处理方法、数据处理装置、电子设备及存储介质 | |
| US20170352162A1 (en) | Region-of-interest extraction device and region-of-interest extraction method | |
| US9633272B2 (en) | Real time object scanning using a mobile phone and cloud-based visual search engine | |
| WO2019056503A1 (zh) | 门店监控评价方法、装置及存储介质 | |
| WO2022007451A1 (zh) | 目标检测方法、装置、计算机可读介质及电子设备 | |
| WO2020001219A1 (zh) | 图像处理方法和装置、存储介质、电子设备 | |
| CN112434612A (zh) | 吸烟检测方法、装置、电子设备及计算机可读存储介质 | |
| CN112199545A (zh) | 基于图片文字定位的关键词显示方法、装置及存储介质 | |
| CN110942456B (zh) | 篡改图像检测方法、装置、设备及存储介质 | |
| CN110909685A (zh) | 姿势估计方法、装置、设备及存储介质 | |
| CN113989336A (zh) | 一种可见光图像和红外图像配准方法及装置 | |
| Mu et al. | Finding autofocus region in low contrast surveillance images using CNN-based saliency algorithm | |
| CN110610178A (zh) | 图像识别方法、装置、终端及计算机可读存储介质 | |
| US20190149878A1 (en) | Determining and correlating visual context on a user device with user behavior using digital content on the user device | |
| CN115063473A (zh) | 物体高度的检测方法、装置、计算机设备、存储介质 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18904410 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 18904410 Country of ref document: EP Kind code of ref document: A1 |