Summary of the invention
This specification one or more embodiment describes a kind of method and apparatus of controlling terminal movement, claps without artificial
Scene photograph and setting loss photo are taken the photograph, so as to shorten the Claims Resolution period.
In a first aspect, providing a kind of method of controlling terminal movement, method includes:
It is spaced to schedule, obtains at least one image of the terminal acquisition;
According at least one described image, one in multiple predetermined movement directions is determined using neural network classification model
Predetermined movement direction is target direction of motion;
It controls the terminal and moves default unit according to the target direction of motion, so that the terminal is held along motion profile
Continuous acquisition image, described image are used for car damage identification.
In a kind of possible embodiment, the method also includes: it is moved according to the terminal along the motion profile
During continuous collecting multiple images, to vehicle carry out setting loss.
In a kind of possible embodiment, the method also includes:
According to the predetermined time interval, location information and/or posture with the associated terminal of described image are obtained
Information, wherein the location information is the positional relationship information of the terminal and the vehicle, and the posture information is the end
The information of shooting angles at end;
Described at least one image according to, is determined in multiple predetermined movement directions using neural network classification model
One predetermined movement direction is target direction of motion, comprising:
According at least one described image, and location information and/or posture with the associated terminal of described image
Information determines that a predetermined movement direction in multiple predetermined movement directions is target movement side using neural network classification model
To.
In a kind of possible embodiment, the method also includes:
According to the predetermined time interval, location information and/or posture with the associated terminal of described image are obtained
Information, wherein the location information is the positional relationship information of the terminal and the vehicle, and the posture information is the end
The information of shooting angles at end;
According to the terminal along the motion profile move during continuous collecting multiple images, and with the figure
As the location information and/or posture information of the associated terminal, setting loss is carried out to vehicle.
In a kind of possible embodiment, the neural network classification model is based on training sample and trains in advance, institute
Stating training sample includes the multiple images under car damage identification scene, and each image has the mark of proven target direction of motion
Label.
In a kind of possible embodiment, the multiple predetermined movement direction includes according to the first preset coordinate system along X
The translation of axis positive axis is translated along the negative semiaxis translation of X-axis, along the translation of Y-axis positive axis, along the negative semiaxis translation of Y-axis, along Z axis positive axis
With along at least one of Z axis negative semiaxis translation;
It controls the terminal and moves default unit according to the target direction of motion, comprising:
It controls the terminal and translates pre-determined distance according to the target direction of motion.
Further, first preset coordinate system sets according to the position of the vehicle.
In a kind of possible embodiment, the multiple predetermined movement direction includes according to the second preset coordinate system around X
Axis rotates clockwise, rotates counterclockwise around X-axis, rotating clockwise around Y-axis, rotate counterclockwise around Y-axis, rotate clockwise about the z axis,
At least one of rotate counterclockwise about the z axis;
It controls the terminal and moves default unit according to the target direction of motion, comprising:
It controls the terminal and rotates predetermined angle according to the target direction of motion.
Further, second preset coordinate system sets according to the position of the terminal.
In a kind of possible embodiment, the control terminal moves default single according to the target direction of motion
Position, comprising:
It is moved by terminal described in unmanned plane, automatic running robot or manipulator control according to the target direction of motion
Default unit.
Second aspect, provides a kind of device of controlling terminal movement, and device includes:
Acquiring unit obtains at least one image of the terminal acquisition for being spaced to schedule;
Determination unit, at least one image for being obtained according to the acquiring unit, uses neural network classification model
Determine that a predetermined movement direction in multiple predetermined movement directions is target direction of motion;
Control unit, it is default single for controlling the target direction of motion movement that the terminal is determined according to the determination unit
Position, so that the terminal, along motion profile continuous collecting image, described image is used for car damage identification.
The third aspect provides a kind of computer readable storage medium, is stored thereon with computer program, when the calculating
When machine program executes in a computer, enable computer execute first aspect method.
Fourth aspect provides a kind of calculating equipment, including memory and processor, and being stored in the memory can hold
Line code, when the processor executes the executable code, the method for realizing first aspect.
The method and apparatus provided by this specification embodiment obtain terminal acquisition first, in accordance with predetermined time interval
At least one image determine multiple predetermined movements using neural network classification model then according at least one described image
A predetermined movement direction in direction is target direction of motion, then controls the terminal and move according to the target direction of motion
Default unit, so that the terminal, along motion profile continuous collecting image, described image is used for car damage identification.Therefore base
In at least one acquired image of terminal, can determine how that terminal is made to move to next position from current location, from
And determine terminal motion profile so that terminal along the motion profile move during the image that acquires meet and survey the need of setting loss
It asks, is not necessarily to artificial floor photo and setting loss photo, so as to shorten the Claims Resolution period.
Specific embodiment
With reference to the accompanying drawing, the scheme provided this specification is described.
Fig. 1 is the implement scene schematic diagram of one embodiment that this specification discloses.As shown in Figure 1, vehicle 11 is to need
Survey the vehicle of setting loss, insurance company sends the equipment 12 for carrying camera to come car accident scene, acquisition vehicle 11
Image, wherein the equipment 12 of above-mentioned carrying camera can be, but not limited to as unmanned plane, automatic running robot or manipulator etc.
Automatic equipment.In this specification embodiment, the action of the equipment 12 of camera can be carried by automatic path planning algorithmic rule
Path to find damage location, and acquires the photo and/or video recording of damage location.Optionally, it can also acquire and carry camera
Equipment 12 is in the motion track information in above-mentioned action path motion process, such as camera posture information, camera and vehicle position
Relation information etc. is set, and camera photos and video information are saved, image algorithm can be used above- mentioned information and carry out to vehicle
Setting loss.
Mainly include following treatment process: 1, path planning in scene shown in FIG. 1: being received according to the equipment 12 for carrying camera
Location information, posture information and the image information of collection to carry camera the next direction of motion of equipment 12 plan, example
Such as, the equipment 12 for carrying camera is planned from the action path that position A moves to position B in Fig. 1.2, it Image Acquisition: takes
The camera continuous collecting and record image information of equipment 12 with camera, it is to be understood that camera can be recorded a video with continuous collecting
(i.e. video), camera can also acquire photo, such as acquire one or more image in position A, in position B acquire again one or
Multiple images.3, image recognition: in conjunction with routing information (for example, camera posture information, camera and vehicle location relation information etc.)
Car damage identification is carried out using image recognition technology with image information.As an example, can be according to camera posture information, camera and vehicle
Positional relationship information and image information summarize the component and degree for carrying out identification of damage to setting loss algorithm, and determine and repair price.
In this specification embodiment, above-mentioned camera is referred to as terminal, that is to say, that above-mentioned camera can be to be exclusively used in
The special equipment taken pictures or recorded a video, or be applied not only to take pictures or record a video also general with communication function or processing function
Equipment, above-mentioned common apparatus such as mobile phone, plate etc..
It should be noted that can only acquire image information in above-mentioned scene, camera posture letter can also be further acquired
The routing informations such as breath, camera and vehicle location relation information correspondingly, can be only in path planning or image recognition processes
It, can not also be only in accordance with image information herein in connection with camera posture information, camera and vehicle location relation information according to image information
Equal routing informations, so as to form a variety of feasible schemes, these schemes are in the scheme of this specification embodiment offer
Within the scope of.
Fig. 2 shows the method flow diagrams moved according to the controlling terminal of one embodiment.The executing subject of this method can be with
It is the equipment 12 shown in FIG. 1 for carrying camera (i.e. terminal), or the control system in addition set up, the control system can be with
It is communicated with the equipment 12 for carrying camera, so that how the equipment 12 for controlling carrying camera moves.As shown in Fig. 2, the implementation
The method of controlling terminal movement obtains the terminal acquisition the following steps are included: step 21, is spaced to schedule in example
At least one image;Step 22, according at least one described image, multiple predetermined movements are determined using neural network classification model
A predetermined movement direction in direction is target direction of motion;Step 23, the terminal is controlled according to the target movement side
To default unit is moved, so that the terminal, along motion profile continuous collecting image, described image is used for car damage identification.It retouches below
State the specific executive mode of above each step.
It first in step 21, is spaced to schedule, obtains at least one image of the terminal acquisition.It is understood that
, during the terminal is moved along the motion profile, continuous collecting image, wherein the mode for acquiring image can be with
For shooting photo or video recording.In one example, it during the terminal is moved along the motion profile, is persistently recorded
Picture, at least one above-mentioned image are the video frame extracted in the video recording;In another example, the terminal is along the fortune
During dynamic rail mark moves, a photo is shot at interval of preset time, at least one above-mentioned image is that terminal is shot extremely
A few photo.
In addition, it is necessary to illustrate, at least one image in step 21 is used for path planning, the number of at least one image
Mesh can be one or multiple, when the number of at least one above-mentioned image is multiple, this multiple images can for
The multiple images of same position acquisition, or in the multiple images of different location acquisition.
Optionally, the position with the associated terminal of described image can also be obtained according to the predetermined time interval
Information and/or posture information, wherein the location information is the positional relationship information of the terminal and the vehicle, the appearance
State information is the information of shooting angles of the terminal.
Then in step 22, according at least one described image, multiple default fortune are determined using neural network classification model
A predetermined movement direction in dynamic direction is target direction of motion.Optionally, the position of the terminal is also obtained in step 21
Confidence breath and/or posture information can correspondingly close in step 22 according at least one described image, and with described image
The location information and/or posture information of the terminal of connection, determine multiple predetermined movement directions using neural network classification model
In a predetermined movement direction be target direction of motion.
It is understood that the neural network classification model is based on training sample and trains in advance, the training sample
Including the multiple images under car damage identification scene, each image has the label of proven target direction of motion.
In one example, the multiple predetermined movement direction includes flat along X-axis positive axis according to the first preset coordinate system
It moves, translated along the negative semiaxis translation of X-axis, along the translation of Y-axis positive axis, along the negative semiaxis translation of Y-axis, along Z axis positive axis and along Z axis negative half
At least one of axis translation.
Wherein, first preset coordinate system can preset, such as the position according to the vehicle is set, and Fig. 3 is
This specification disclose one embodiment the first preset coordinate system schematic diagram, as shown in figure 3, the first preset coordinate system with
It is established centered on vehicle location.
In one example, the multiple predetermined movement direction includes turning clockwise according to the second preset coordinate system around X-axis
It is dynamic, rotated counterclockwise around X-axis, rotated clockwise around Y-axis, rotate counterclockwise around Y-axis, rotate clockwise about the z axis, inverse time about the z axis
At least one of needle rotation.
Wherein, second preset coordinate system can change with the position change of terminal, such as according to the terminal
Position setting, Fig. 4 are the schematic diagram of the second preset coordinate system of one embodiment that this specification discloses, as shown in figure 4, this
Two preset coordinate systems are established centered on camera (i.e. terminal) position.
Finally in step 23, the terminal is controlled according to the target direction of motion and moves default unit, so that the end
End is used for car damage identification along motion profile continuous collecting image, described image.It is understood that the every movement of terminal is primary, shape
At one section of motion profile of terminal, terminal can be by repeatedly movement, and the componental movement track for the terminal that movement is formed every time is total
With the entire motion track for constituting terminal.
In one example, the movement of terminal may include translation and rotation.When target direction of motion belongs to translation, then
It controls the terminal and translates pre-determined distance according to the target direction of motion.When target direction of motion belongs to rotation, then control
The terminal rotates predetermined angle according to the target direction of motion.
Optionally, after step 23, can according to the terminal along the motion profile move during persistently adopt
The multiple images of collection carry out setting loss to vehicle.It is understood that carrying out setting loss to vehicle and determining that terminal motion profile can be with
It is executed, can also be executed by different equipment, this specification embodiment is not limited this by same equipment.Wherein it is possible to only
According to the image of acquisition, setting loss is carried out to vehicle.Alternatively, according to the terminal along the motion profile move during continue
The multiple images of acquisition, and location information and/or posture information with the associated terminal of described image carry out vehicle
Setting loss.
In one example, carrying out setting loss to vehicle can include determining that in defective component, degree of injury and reparation price
It is one or more.
In one example, one or more neural network classification models can be taken to carry out setting loss to vehicle, herein not
It repeats.
Furthermore, it is possible to be transported by terminal described in unmanned plane, automatic running robot or manipulator control according to the target
Dynamic direction moves default unit.
The method provided by this specification embodiment obtains terminal acquisition at least first, in accordance with predetermined time interval
One image is determined in multiple predetermined movement directions then according at least one described image using neural network classification model
A predetermined movement direction be target direction of motion, then control the terminal moved according to the target direction of motion it is default single
Position, so that the terminal, along motion profile continuous collecting image, described image is used for car damage identification.Therefore it is based on terminal
At least one acquired image can determine how that terminal is made to move to next position from current location, so that it is determined that
Terminal motion profile so that terminal along the motion profile move during the image that acquires meet the needs of surveying setting loss, nothing
Artificial floor photo and setting loss photo are needed, so as to shorten the Claims Resolution period.
According to the embodiment of another aspect, a kind of device of controlling terminal movement is also provided.Fig. 5 is shown to be implemented according to one
The schematic block diagram of the device of the controlling terminal movement of example.As shown in figure 5, the device 500 includes:
Acquiring unit 51 obtains at least one image of the terminal acquisition for being spaced to schedule;
Determination unit 52, at least one image for being obtained according to the acquiring unit 51, uses neural network classification
Model determines that a predetermined movement direction in multiple predetermined movement directions is target direction of motion;
Control unit 53 moves in advance for controlling the terminal according to the target direction of motion that the determination unit 52 determines
If unit, so that the terminal, along motion profile continuous collecting image, described image is used for car damage identification.
In one example, described device further include:
Setting loss unit, for according to the terminal along described control unit 53 determine motion profile move during hold
The multiple images of continuous acquisition carry out setting loss to vehicle.
In one example, the acquiring unit 51 is also used to according to the predetermined time interval, acquisition and described image
The location information and/or posture information of the associated terminal, wherein the location information is the terminal and the vehicle
Positional relationship information, the posture information are the information of shooting angles of the terminal;
The determination unit 52, specifically at least one image obtained according to the acquiring unit 51, and with institute
The location information and/or posture information for stating the associated terminal of image are determined multiple default using neural network classification model
A predetermined movement direction in the direction of motion is target direction of motion.
In one example, the acquiring unit 51 is also used to according to the predetermined time interval, acquisition and described image
The location information and/or posture information of the associated terminal, wherein the location information is the terminal and the vehicle
Positional relationship information, the posture information are the information of shooting angles of the terminal;
Described device further include:
Setting loss unit, for according to the terminal along described control unit 53 determine motion profile move during hold
The multiple images of continuous acquisition, and location information and/or posture information with the associated terminal of described image, to vehicle into
Row setting loss.
In one example, the neural network classification model is based on training sample and trains in advance, the training sample
Including the multiple images under car damage identification scene, each image has the label of proven target direction of motion.
In one example, the multiple predetermined movement direction includes flat along X-axis positive axis according to the first preset coordinate system
It moves, translated along the negative semiaxis translation of X-axis, along the translation of Y-axis positive axis, along the negative semiaxis translation of Y-axis, along Z axis positive axis and along Z axis negative half
At least one of axis translation;
Described control unit 53, the target movement side determined specifically for controlling the terminal according to the determination unit 52
To translation pre-determined distance.
Further, first preset coordinate system sets according to the position of the vehicle.
In one example, the multiple predetermined movement direction includes turning clockwise according to the second preset coordinate system around X-axis
It is dynamic, rotated counterclockwise around X-axis, rotated clockwise around Y-axis, rotate counterclockwise around Y-axis, rotate clockwise about the z axis, inverse time about the z axis
At least one of needle rotation;
Described control unit 53, the target movement side determined specifically for controlling the terminal according to the determination unit 52
To rotation predetermined angle.
Further, second preset coordinate system sets according to the position of the terminal.
In one example, described control unit 53 are specifically used for passing through unmanned plane, automatic running robot or manipulator
It controls the terminal and moves default unit according to the target direction of motion that the determination unit 52 determines.
The device provided by this specification embodiment is spaced by acquiring unit 51 to schedule first, obtains institute
At least one image of terminal acquisition is stated, neural network point is then used according at least one described image by determination unit 52
Class model determines that a predetermined movement direction in multiple predetermined movement directions is target direction of motion, then is controlled by control unit 53
It makes the terminal and moves default unit according to the target direction of motion, so that the terminal is along motion profile continuous collecting figure
Picture, described image are used for car damage identification.Therefore at least one image acquired based on terminal, it can determine how
Terminal is set to move to next position from current location, so that it is determined that terminal motion profile, so that terminal is transported along the motion profile
The image acquired in dynamic process meets the needs of surveying setting loss, artificial floor photo and setting loss photo is not necessarily to, to contract
The short Claims Resolution period.
According to the embodiment of another aspect, a kind of computer readable storage medium is also provided, is stored thereon with computer journey
Sequence enables computer execute method described in Fig. 2 when the computer program executes in a computer.
According to the embodiment of another further aspect, a kind of calculating equipment, including memory and processor, the memory are also provided
In be stored with executable code, when the processor executes the executable code, realize
Method described in Fig. 2.
Those skilled in the art are it will be appreciated that in said one or multiple examples, function described in the invention
It can be realized with hardware, software, firmware or their any combination.It when implemented in software, can be by these functions
Storage in computer-readable medium or as on computer-readable medium one or more instructions or code transmitted.
Above-described specific embodiment has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that being not intended to limit the present invention the foregoing is merely a specific embodiment of the invention
Protection scope, all any modification, equivalent substitution, improvement and etc. on the basis of technical solution of the present invention, done should all
Including within protection scope of the present invention.