CN109934949A

CN109934949A - Work attendance method and device, equipment, storage medium

Info

Publication number: CN109934949A
Application number: CN201910185795.0A
Authority: CN
Inventors: 吴佳飞; 丁永超; 汪文轩; 刘小峰; 金古; 张广程
Original assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Current assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Priority date: 2019-03-12
Filing date: 2019-03-12
Publication date: 2019-06-25

Abstract

The embodiment of the present application discloses Work attendance method and device, equipment, storage medium, wherein the described method includes: an acquisition at least frame waits for attendance image；Adaptive partition processing is carried out to attendance image to described, obtains the target area set for meeting predetermined identification condition；Each target object in the set of the target area is identified, is obtained with described to the corresponding identity information set of attendance image；Based on preset object set and each to the corresponding identity information set of attendance image, checking-in result is generated.

Description

Work attendance method and device, equipment, storage medium

Technical field

The invention relates to computer vision techniques, relate to, but are not limited to Work attendance method and device, equipment, storage Jie Matter.

Background technique

During the teaching and work management of school and enterprises and institutions, roll-call is an important content.Tradition Artificial roll-call mode have efficiency is too low, teacher workload is big, spends the time excessive to be not easy with the data of conventional point name record The defects of utilization.Currently based on the attendance checking system of Face datection, for example, teacher at school when, take pictures, obtain to classroom situation To attendance image；Then, this is waited for that attendance image is uploaded to attendance checking system, so that attendance checking system waits in attendance image this Face is identified, to achieve the purpose that realize non-inductive roll-call.Since the system is taken pictures without student to specified region, Roll-call can be realized, thus the roll-call time being greatly saved on classroom.However, based on to the implementation roll-call of attendance image When, there is a problem of that roll-call accuracy is lower.

Summary of the invention

In view of this, the embodiment of the present application is to solve the problems, such as present in the relevant technologies at least one and provide Work attendance method And device, equipment, storage medium.

The technical solution of the embodiment of the present application is achieved in that

In a first aspect, the embodiment of the present application provides a kind of Work attendance method, which comprises an acquisition at least frame waits for attendance Image；Adaptive partition processing is carried out to attendance image to described, obtains the target area set for meeting predetermined identification condition；It is right Each target object in the set of the target area identified, is obtained with described to the corresponding set of identity information of attendance image It closes；Based on preset object set and each to the corresponding identity information set of attendance image, checking-in result is generated.

In other embodiments, described to carry out adaptive partition processing to attendance image to described, it obtains meeting predetermined knowledge Gather the target area of other condition, comprising: carry out subregion to attendance image to described, obtain set of image regions, described image Image-region in regional ensemble includes at least a target object；Image-region in described image regional ensemble is put Greatly, the target area set for meeting predetermined identification condition is obtained.

In other embodiments, described to carry out subregion to attendance image to described, obtain set of image regions, comprising: root According to it is described to each target object in attendance image in the position on attendance image, divide to attendance image described Area obtains set of image regions.

In other embodiments, the image-region in described image regional ensemble amplifies, and obtains meeting pre- Surely the target area set of condition is identified, comprising: be amplified to the size of the target object each in described image region pre- If size, target area set is obtained.

In other embodiments, it is described according to each target object in attendance image described on attendance image Position, carry out subregion to attendance image to described, obtain set of image regions, comprising: exist based on each target object The position on attendance image, determines the size of the target object；Based on the size of each target object, to institute It states and carries out subregion to attendance image, obtain set of image regions.

In other embodiments, the image-region in described image regional ensemble amplifies, and obtains meeting pre- Surely the target area set of condition is identified, comprising: the size based on each target object in each described image region determines Amplification factor corresponding with described image region；Based on the amplification factor in each described image region, to the amplification factor Corresponding image-region amplifies, and obtains target area set.

In other embodiments, the size based on each target object in each described image region, determine with The corresponding amplification factor in described image region, comprising: the size based on each target object in described image region determines institute State the average-size of the target object in image-region；Based on the average-size of the target object in described image region, from institute It states and determines benchmark image region in set of image regions；The average-size for determining the target object in the benchmark image region, with Ratio in described image region between the average-size of target object；Based on the ratio, putting for described image region is determined Big multiple.

In other embodiments, an acquisition at least frame waits for attendance image, comprising: by the initial acquisition to attendance image Moment to the duration between current time, is determined as attendance duration；Based on each to the corresponding identity information set of attendance image, Determine target object number；If the attendance duration is less than preset duration, and the target object number is less than preset number, It resurveys an at least frame and waits for attendance image.

In other embodiments, described based on preset object set and each to the corresponding set of identity information of attendance image It closes, generates checking-in result, comprising: if the attendance duration is less than preset duration, and the target object number is equal to default Number is based on preset object set and each to the corresponding identity information set of attendance image, generates checking-in result；If institute Attendance duration is stated more than or equal to preset duration, based on preset object set and each to the corresponding set of identity information of attendance image It closes, generates checking-in result.

In other embodiments, the method also includes: the checking-in result is sent to management platform.

In other embodiments, each target object in the set of the target area identifies, obtains To with described to the corresponding identity information set of attendance image, comprising: to each target in the set of the target area Object is identified, fisrt feature data corresponding with the target object are obtained；First based on each target object Characteristic and preset property data base determine identity information corresponding with the target object, obtain described to attendance figure As corresponding identity information set.

In other embodiments, the fisrt feature data and preset characteristic based on each target object Library determines identity information corresponding with the target object, obtains described to the corresponding identity information set of attendance image, packet It includes: from preset property data base, the determining second feature data to match with each fisrt feature data；Based on each phase Matched second feature data obtain identity information corresponding with the second feature data from the property data base, obtain To described to the corresponding identity information set of attendance image.

In other embodiments, before at least a frame waits for attendance image for the acquisition, the method also includes: it is loaded into institute State the feature image of each attendance object recorded on object set, the title and identity of each attendance object；Wherein, institute Identity is stated for attendance object described in unique identification；Each feature image is identified, is obtained and the attendance The corresponding second feature data of object；The title, second feature data and identity of each attendance object are stored in default Property data base in.

In other embodiments, described based on each second feature data to match, it is obtained from the property data base Identity information corresponding with the second feature data is taken, is obtained with described to the corresponding identity information set of attendance image, packet Include: based on each second feature data to match, from identity documented by the property data base, obtain with it is described The corresponding target identities mark of second feature data；It is identified based on each target identities, is obtained from the property data base Take target object title corresponding with target identities mark；What be will acquire is described to each target object in attendance image Target identities mark and target object title, are determined as identity information corresponding with the target object, obtain needing checking with described The corresponding identity information set of diligent image.

In other embodiments, described to be stored in the title, second feature data and identity of each attendance object In preset property data base, comprising: by the second feature data and identity of each attendance object, be stored in the feature In database；Using each identity as key assignments, each key assignments and attendance object oriented corresponding with the key assignments are write Enter Hash table, the Hash table is stored in the property data base.

In other embodiments, described based on each target identities mark, obtained from the property data base with The target identities identify corresponding target object title, comprising: key assignments are identified as with each target identities, described in access Hash table obtains target object title corresponding with target identities mark.

In other embodiments, the target object is face.

Second aspect, the embodiment of the present application provide a kind of Work attendance device, comprising: image capture module is configured to acquisition extremely A few frame waits for attendance image；Image processing module is configured to carry out adaptive partition processing to attendance image to described, be accorded with Close the target area set of predetermined identification condition；Object Identification Module, each mesh being configured in gathering the target area Mark object is identified, is obtained with described to the corresponding identity information set of attendance image；Checking-in result generation module, configuration Are as follows: it is based on preset object set and each to the corresponding identity information set of attendance image, generates checking-in result.

In other embodiments, described image processing module, comprising: multidomain treat-ment submodule is configured that and needs checking described Diligent image carries out subregion, obtains set of image regions, and the image-region in described image regional ensemble includes at least a target Object；Enhanced processing submodule is configured that and amplifies to the image-region in described image regional ensemble, obtains meeting pre- Surely the target area set of condition is identified.

In other embodiments, the multidomain treat-ment submodule is configured that according to described to each target in attendance image Object carries out subregion to attendance image in the position on attendance image, to described, obtains set of image regions.

In other embodiments, the enhanced processing submodule is configured that the target each in described image region The size of object is amplified to pre-set dimension, obtains target area set.

In other embodiments, the multidomain treat-ment submodule includes: size determination unit, is configured that based on each institute Target object is stated in the position on attendance image, determines the size of the target object；Multidomain treat-ment unit, configuration Are as follows: the size based on each target object carries out subregion to attendance image to described, obtains set of image regions.

In other embodiments, the enhanced processing submodule includes: amplification factor determination unit, is configured that based on every The size of each target object in one described image region determines amplification factor corresponding with described image region；At amplification Manage unit, be configured that the amplification factor based on each described image region, to image-region corresponding with the amplification factor into Row amplification obtains target area set.

In other embodiments, the amplification factor determination unit is configured that based on each mesh in described image region The size for marking object, determines the average-size of the target object in described image region；Based on the target in described image region The average-size of object determines benchmark image region from described image regional ensemble；Determine the mesh in the benchmark image region Mark the average-size of object, the ratio between the average-size of target object in described image region；Based on the ratio, really Determine the amplification factor in described image region.

In other embodiments, described image acquisition module be configured that by the initial acquisition moment of attendance image to working as Duration between the preceding moment is determined as attendance duration；Based on each to the corresponding identity information set of attendance image, target is determined Object number；If the attendance duration is less than preset duration, and the target object number is less than preset number, resurveys At least a frame waits for attendance image.

In other embodiments, the checking-in result generation module is configured that if when the attendance duration is less than default It is long, and the target object number is equal to preset number, based on preset object set and each to the corresponding body of attendance image Part information aggregate, generates checking-in result；If the attendance duration is more than or equal to preset duration, based on preset object set and It is each to the corresponding identity information set of attendance image, generate checking-in result.

In other embodiments, described device further include: sending module is configured to the checking-in result being sent to management Platform.

In other embodiments, the Object Identification Module, comprising: Object identifying submodule is configured that the target Each target object in regional ensemble is identified, fisrt feature data corresponding with the target object are obtained；Body Part determines submodule, is configured that the fisrt feature data based on each target object and preset property data base, determines Identity information corresponding with the target object obtains described to the corresponding identity information set of attendance image.

In other embodiments, the identity determines that submodule includes: characteristics determining unit, is configured that from preset spy It levies in database, the determining second feature data to match with each fisrt feature data；Identity determination unit is configured that base In each second feature data to match, identity corresponding with the second feature data is obtained from the property data base Information obtains described to the corresponding identity information set of attendance image.

In other embodiments, described device further includes that information insmods and information storage module；Wherein, the information It insmods, is configured that the feature image, each attendance object for being loaded into each attendance object recorded on the object set Title and identity；Wherein, the identity is for attendance object described in unique identification；The Object Identification Module, matches It is set to: each feature image is identified, obtain second feature data corresponding with the attendance object；The information Memory module is configured that the title, second feature data and identity of each attendance object being stored in preset characteristic According in library.

In other embodiments, the identity determination unit, comprising: identity obtains subelement, is configured to every The one second feature data to match obtain and the second feature from identity documented by the property data base The corresponding target identities mark of data；Object oriented obtains subelement, is configured that and is identified based on each target identities, from Target object title corresponding with target identities mark is obtained in the property data base；Identity information determines subelement, It is configured that the target identities mark and target object title to each target object in attendance image that will acquire, determines For identity information corresponding with the target object, obtain with described to the corresponding identity information set of attendance image.

In other embodiments, the information storage module be configured that the second feature data of each attendance object and Identity is stored in the property data base；Using each identity as key assignments, by each key assignments and with the key It is worth corresponding attendance object oriented write-in Hash table, the Hash table is stored in the property data base.

In other embodiments, the object oriented obtains subelement, is configured that and is identified as with each target identities Key assignments accesses the Hash table, obtains target object title corresponding with target identities mark.

In other embodiments, the target object is face.

The third aspect, the embodiment of the present application provide a kind of electronic equipment, including memory and processor, the memory are deposited The computer program that can be run on a processor is contained, the processor is realized in above-mentioned Work attendance method when executing described program Step.

Fourth aspect, the embodiment of the present application provide a kind of computer readable storage medium, are stored thereon with computer program, The computer program realizes the step in above-mentioned Work attendance method when being executed by processor.

In the embodiment of the present application, adaptive partition processing can be carried out to attendance image to acquisition, obtain meeting predetermined The target area of identification condition is gathered, in this way, when being identified to each target object in the set of the target area, energy Each target object is enough accurately identified, to improve attendance accuracy.

Detailed description of the invention

Fig. 1 is the implementation process schematic diagram of the embodiment of the present application Work attendance method；

Fig. 2 is the subregion schematic diagram that the embodiment of the present application waits for attendance image；

Fig. 3 is that the embodiment of the present application waits for attendance picture material schematic diagram；

Fig. 4 is the implementation process schematic diagram of another Work attendance method of the embodiment of the present application；

Fig. 5 is the composed structure schematic diagram of the non-inductive check class attendance device of the embodiment of the present application；

Fig. 6 is the embodiment of the present application to classroom panoramic picture and sectional image schematic diagram；

Fig. 7 is the implementation process schematic diagram of the another Work attendance method of the embodiment of the present application；

Fig. 8 is the implementation process schematic diagram of the embodiment of the present application adaptive-interpolation strategy；

Fig. 9 is the composed structure schematic diagram of the embodiment of the present application Work attendance device；

Figure 10 is a kind of hardware entities schematic diagram of the embodiment of the present application electronic equipment.

Specific embodiment

The technical solution of the application is further elaborated on reference to the accompanying drawings and examples.

The embodiment of the present application provides a kind of Work attendance method, and this method is applied to electronic equipment, the function that this method is realized It can be realized by the processor caller code in electronic equipment, certain program code can be stored in computer storage In medium, it is seen then that the electronic equipment includes at least pocessor and storage media.

Fig. 1 is the implementation process schematic diagram of the embodiment of the present application Work attendance method, as shown in Figure 1, the method comprising the steps of S101 to step S104:

S101, an acquisition at least frame wait for attendance image；

It is to be appreciated that by attendance object may be dynamic change in the scene having, a frame is only acquired in this way and is waited for Attendance image can not capture all target objects of hall completely.Such as when classroom is called the roll, have on classroom Raw current low head, some students currently lift head, and the frame acquired at this time waits for not capturing in attendance image all The face feature of student.Therefore, in practical applications, attendance image can be waited for acquire multiframe by way of shooting video, Can also multiframe be acquired and wait for attendance image according to the preset time interval.In this way, attendance image is waited for by acquiring an at least frame, Attendance image is waited for implement to call the roll using the multiframe of acquisition, can be improved roll-call accuracy.

S102, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S103, each target object in the set of the target area is identified, is obtained with described to attendance image Corresponding identity information set；

In the present embodiment, when the target object is face, Work attendance method provided by the embodiment of the present application is usually used in The scenes such as check class attendance, meeting-place attendance, examination hall attendance.Certainly, in other embodiments, the target object can also be dynamic Object does not limit the object type of the target object here.In practical applications, it can be set by the electronics with attendance ability Standby control camera is described to attendance image to acquire.In order to clearly and completely collect each mesh in attendance region Object is marked, and does not need target object cooperation, it is generally the case that camera is vacantly mounted on to the front in attendance region, with Make each target object in attendance image include attendance region collected.

In general, it is shot to attendance image by being mounted on the video camera of fixed viewpoint, and fixed viewpoint must Near big and far smaller picture characteristics will be present, it is therefore desirable to which the region division of adaptive progress image carries out figure to different zones The pretreatment of picture, so that the target object in each region after dividing meets identification condition.For example, being a religion to attendance image Room is attended class the photo of situation, and the facial size of front-seat student is far longer than the size of last a few row's faces in photo, due to last The facial size of several rows is smaller, when causing to carry out recognition of face to the photo, it is difficult to the accurate face characteristic for extracting last several rows, So as to cause attendance accuracy decline.Therefore, in the embodiment of the present application, known in the target object treated in attendance image It before not, first treats attendance image and carries out adaptive partition processing, so that each target object symbol in attendance image It closes identification condition (for example, each target object is made to meet pre-set dimension), obtains target area set, then, then to mesh Each target object in mark regional ensemble is identified, target object title is obtained.For example, the target object is face When, the target object title is name.

It should be noted that treating attendance image carries out adaptive partition processing, in obtained target area set at least Including a target area, a target object is included at least in each target area.

S104, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

For example, the object set can be class's list in the application scenarios that classroom is called the roll.The checking-in result can To include name and student number, the name of personnel absent from duty, student number, the number absent from duty etc. of the personnel of registering.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S201 to step S205:

S201, an acquisition at least frame wait for attendance image；

S202, subregion is carried out to attendance image to described, obtains set of image regions, the figure in described image regional ensemble As region includes at least a target object；

It should be noted that two adjacent image-regions have partial region when treating attendance image progress subregion Content is identical, the picture material this makes it possible to each target object in the set of image regions ensured be all it is complete, It is subsequent based on target area set to make, it, can be accurate when being identified to each target object in each target area Identification is to each target object in attendance image.For example, shown in Fig. 2, to include target object 201 in attendance image 20 to mesh Mark object 204, wherein target object 201 and target object 202 are divided in image-region 21, target object 203 and target Object 204 is divided in image-region 22, wherein the sub-district in subregion 211 and image-region 22 in image-region 21 The content in domain 221 is identical.

S203, the image-region in described image regional ensemble is amplified, obtains the mesh for meeting predetermined identification condition Mark regional ensemble；

It is to be appreciated that since the relative position between multiple target objects is different, to there are sizes in attendance image not One target object, therefore, in order to targetedly be pre-processed to target object, can first treat attendance image into Row subregion, obtains set of image regions；Then, the image-region that predetermined amplification condition is met in set of image regions is put Greatly, so that the target area set for meeting predetermined identification condition is obtained, for example, being only less than preset threshold to the size of target object Image-region amplify, can reduce image processing load in this way, can be by image district to be amplified when amplifying Domain is amplified to pre-set dimension, to obtain the target area set for meeting predetermined identification condition.

Here, it should be noted that step S202 and step S203 is actually one of step S102 in above-described embodiment Kind implements example.

S204, each target object in the set of the target area is identified, is obtained described to attendance image pair The identity information set answered；

S205, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

In the embodiment of the present application, the image-region obtained after subregion can be amplified, makes amplified image district Each target object in domain (i.e. target area) meets identification condition, to improve when identifying to each target object Recognition accuracy.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S301 to step S305:

S301, an acquisition at least frame wait for attendance image；

To each target object in attendance image in the position on attendance image described in S302, basis, to described Subregion is carried out to attendance image, obtains set of image regions；

In general, target object is referring to target object institute on to attendance image to the position on attendance image The region that accounts for, size.Can based on to each target object in attendance image in the size on attendance image, to described Subregion is carried out to attendance image, obtains set of image regions.

For example, including target object 301 to target object 309 to attendance image 30, wherein target object shown in Fig. 3 301 and target object 302 be greater than target object 303 and target object 304 to attendance image to the size in attendance image In size be greater than target object 305 and target object 306 and be greater than target object 307 to target to the size in attendance image Object 309 is to the size in attendance image.That is, target object 301 and target object 302 be within the scope of first size, target Object 303 and target object 304 are in the second size range, and target object 305 and target object 306 are in third size range Interior, target object 307 to target object 309 is in the 4th size range, therefore, when treating the progress subregion of attendance image 30, Target object 301 and target object 302 can be divided in image-region 31, target object 303 and target object 304 are drawn Point in image-region 32, target object 305 and target object 306 are divided in image-region 33, by target object 307 to Target object 309 is divided in image-region 34, and obtained set of image regions includes image-region 31 to image-region 34.

Here, it should be noted that step S302 is actually a kind of implementation example of step S202 in above-described embodiment.

S303, the image-region in described image regional ensemble is amplified, obtains the mesh for meeting predetermined identification condition Mark regional ensemble；

S304, each target object in the set of the target area is identified, is obtained described to attendance image pair The identity information set answered；

S305, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

In the embodiment of the present application, can based on to each target object in attendance image in the position on attendance image It sets, carries out subregion to attendance image to described, in this way, all include complete target object in obtained each image-region, from And improve the recognition accuracy of target object.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S401 to step S405:

S401, an acquisition at least frame wait for attendance image；

To each target object in attendance image in the position on attendance image described in S402, basis, to described Subregion is carried out to attendance image, obtains set of image regions；

S403, the size of the target object each on the image-region in described image regional ensemble is amplified to it is default Size obtains target area set；

In practical applications, the pre-set dimension can be arranged in user according to demand, and the electronic equipment can be by image The target object that target object is less than pre-set dimension in region amplifies.For example, shown in Fig. 3, wherein image-region 31 and figure As the size of the target object in region 32 is all larger than pre-set dimension, image-region 33 and the target object in image-region 34 Size is respectively less than pre-set dimension, therefore, can only amplify to image-region 33 and image-region 34, so that image-region 33 In target object 305 and target object 306, image-region 34 in target object 307 to target object 309 size amplify To pre-set dimension, obtained target area set includes image-region 31, image-region 32, amplified image-region 33 and puts Image-region 34 after big.

Here, it should be noted that step S403 is actually a kind of implementation example of step S303 in above-described embodiment.

In the embodiment of the present application, according to it is described to each target object in attendance image described on attendance image Position carries out subregion to attendance image to described, after obtaining set of image regions, by will be in described image regional ensemble The size of each target object is amplified to pre-set dimension on image-region, each target pair in the target area made As that can be accurately identified.

S404, each target object in the set of the target area is identified, is obtained described to attendance image pair The identity information set answered；

S405, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S501 to step S507:

S501, an acquisition at least frame wait for attendance image；

S502, based on each target object in the position on attendance image, determine the target object Size；

Here, it should be noted that step S502 and step S503 is actually one of step S302 in above-described embodiment Kind implements example.

S503, the size based on each target object carry out subregion to attendance image to described, obtain image-region Set；

It is to be appreciated that the size based on each target object, carries out subregion to attendance image to described, for example, Target object by size in same range is divided into the same area, in this way, after obtaining image-region, convenient for subsequent to every When one image-region carries out corresponding enhanced processing, reduce the complexity of enhanced processing.For example, directly being carried out to image-region slotting Value amplification does not have to carry out further subregion to image-region.

S504, the size based on each target object in each described image region, determining and described image region pair The amplification factor answered；

S505, the amplification factor based on each described image region, to image-region corresponding with the amplification factor into Row amplification obtains target area set；

It is to be appreciated that this size based on each target object in each described image region, determines corresponding diagram As the amplification factor in region, relatively directly, simply.

Here, it should be noted that step S504 and step S505 is actually one of step S303 in above-described embodiment Kind implements example.

S506, each target object in the set of the target area is identified, is obtained described to attendance image pair The identity information set answered；

S507, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S601 to step S610:

S601, an acquisition at least frame wait for attendance image；

S602, based on each target object in the position on attendance image, determine the target object Size；

S603, the size based on each target object carry out subregion to attendance image to described, obtain image-region Set；

S604, the size based on each target object in described image region, determine the target in described image region The average-size of object；

S605, the average-size based on the target object in described image region are determined from described image regional ensemble Benchmark image region；

Under normal conditions, can be using the corresponding image-region of average largest dimensions as benchmark image-region, benchmark image area The amplification factor in domain is 1, that is, does not amplify processing to benchmark image region.

S606, determine the benchmark image region target object average-size, with target pair in described image region Ratio between the average-size of elephant；

S607, it is based on the ratio, determines the amplification factor in described image region；

Still by taking Fig. 3 as an example, step S604 to step S607 is illustrated, it is assumed that the target object of image-region 31 Average-size be s1, s1=(a1+a2)/2, a1 be image-region 31 in target object 301 size, a2 be image-region 31 The size of middle target object 302；The average-size of the target object of image-region 32 is s2, the target object of image-region 33 Average-size is s3, and the average-size of the target object of image-region 34 is s4, s1 > s2 > s3 > s4；It therefore, can be corresponding by s1 Image-region 31 be determined as benchmark image region, determine ratio of the s1 respectively with s2, s3 and s4, then, put according to preset Big multiple mapping table, such as table 1, determine preset range belonging to each ratio, so that it is determined that corresponding amplification factor.

Table 1

Proportional region	Amplification factor
		First preset range: (1,2]	2
Second preset range: (2,3]	3
		Third preset range: (3,4]	4
……	……

Here, it should be noted that step S604 and step S607 is actually one of step S504 in above-described embodiment Kind implements example.

S608, the amplification factor based on each described image region, to image-region corresponding with the amplification factor into Row amplification obtains target area set；

S609, each target object in the set of the target area is identified, is obtained described to attendance image pair The identity information set answered；

S610, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

In the embodiment of the present application, by the average-size of the target object in the determination benchmark image region, with each institute The ratio in image-region between the average-size of target object is stated, and then determines the amplification factor in correspondence image region, to right The image-region answered amplifies, in such manner, it is possible to based on to the relative dimensions between target object in attendance image, to figure As region carries out amplification appropriate.That is, the embodiment of the present application step S604 to step S607 provide it is a kind of simple, can Capable, effective enlargement processing method.

The embodiment of the present application provides another Work attendance method, and Fig. 4 is the realization of the embodiment of the present application another kind Work attendance method Flow diagram, as shown in figure 4, the method comprising the steps of S701 to step S709:

S701, an acquisition at least frame wait for attendance image；

S702, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S703, each target object in the set of the target area is identified, is obtained with described to attendance image Corresponding identity information set；

S704, by the initial acquisition moment of attendance image to the duration between current time, be determined as attendance duration；

It should be noted that acquisition first needs checking when referring to that attendance starts at the initial acquisition moment of attendance image At the time of diligent image corresponds to.

S705, based on each to the corresponding identity information set of attendance image, determine target object number；

S706, determine whether the attendance duration is less than preset duration；If so, executing step S707；Otherwise, step is executed Rapid S708；

S707, determine whether the target object number is less than preset number；If so, returning to step S701；It is no Then, step S708 is executed；

It should be noted that the preset duration and the preset number can be freely arranged in user.It is to be appreciated that such as Attendance duration described in fruit is less than preset duration, and the target object number is less than preset number, needs to resurvey at least one Frame waits for attendance image, continue identification resurvey to the target object in attendance image；This is because in practical applications, The target object in attendance region is typically dynamic, since the embodiment of the present application is a kind of inductionless Work attendance method, i.e., not The object in attendance region is needed to make the required movement of any cooperation attendance, so, one is based only on to attendance image, possibility It can not accurately identify to obtain the title of each target object in attendance region, for example, being that a student attends class feelings to attendance image The photo of condition, if waiting for that attendance image identifies to this based on Face datection, since this waits for the students' affairs division having in attendance image In the state of bowing, so leading to not identify the student.Therefore, under conditions of attendance duration allows, it usually needs acquisition is more It opens to attendance image, completes attendance task.

S708, based on each to the corresponding identity information set of attendance image and preset object set, generate attendance knot Fruit；

It include student name and in class's list for example, including student name and student number in the identity information set Number, then, based on each to the student name of each student, student number and class's name in the corresponding identity information set of attendance image It is single, so that it may to determine which student to class, which student does not arrive class, to generate checking-in result.In order to avoid duplication of name situation Appearance, the checking-in result of generation generally comprises student name and student's student number.

It is to be appreciated that if the attendance duration is less than preset duration, and the target object number is equal to present count Mesh carries out attendance at this time if continuing an acquisition at least frame waits for attendance image, at this time can be based on every also just without meaning One, to the corresponding identity information set of attendance image and preset object set, generates checking-in result.

In addition, if the attendance duration is more than or equal to preset duration, attendance terminates, at this time based on each to attendance figure As corresponding identity information set and preset object set, checking-in result is generated.This is because attendance has certain timeliness Property, for example, if student class is over, resurvey at this time an at least frame wait for attendance image carry out attendance, without Meaning.

S709, the checking-in result is sent to management platform.

After obtaining checking-in result, the checking-in result can be sent to management platform, convenient for unified management and benefit With.For example, being by the educational administration that checking-in result is sent to school when the Work attendance method of the embodiment of the present application is applied to check class attendance System, so as to the reference frame examined as final grade.

In the embodiment of the present application, after obtaining the title to the target object in attendance image, do not directly generate Checking-in result, but pass through attendance duration and target object number, it is determined whether meet the condition for terminating attendance, if discontented Foot, then resurvey an at least frame and wait for attendance image, thus what attendance accuracy caused by avoiding because of target object missing inspection declined Problem.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S801 to step S805:

S801, an acquisition at least frame wait for attendance image；

S802, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S803, each target object in the set of the target area is identified, is obtained and the target pair As corresponding fisrt feature data；

For example, if target object is face, the target area can be gathered in each target area face into Row identification, to extract the corresponding fisrt feature data of each face.

S804, the fisrt feature data based on each target object and preset property data base, it is determining with it is described The corresponding identity information of target object obtains described to the corresponding identity information set of attendance image；

In general, the title and characteristic of each target object, institute are previously stored in preset property data base Can be determined from property data base and fisrt feature data phase based on the fisrt feature data of each target object The characteristic matched, to obtain corresponding target object title.

It should be noted that step S803 and step S804 are any of the above-described as described in the examples " to the target area Each target object in the set of domain is identified, is obtained described to the corresponding identity information set of attendance image " a kind of reality Apply example, that is, step S803 and step S804 is step S103 in above-described embodiment, step S204, step S304, step S404, step S506, step S609 or any one step in step S703 implementation example.

S805, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S901 to step S906:

S901, an acquisition at least frame wait for attendance image；

S902, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S903, each target object in the set of the target area is identified, is obtained and the target pair As corresponding fisrt feature data；

S904, from preset property data base, the determining second feature data to match with each fisrt feature data；

S905, based on each second feature data to match, obtained from the property data base with it is described second special The corresponding identity information of data is levied, is obtained described to the corresponding identity information set of attendance image；

Here, it should be noted that step S904 and step S905 is actually one kind of above-described embodiment step S804 Implement example.

S906, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S1001 to step S1011:

S1001, the feature image that is loaded into each attendance object recorded on preset object set, each attendance object Title and identity；Wherein, the identity is used for the corresponding attendance object of unique identification；

For example, the object set is the student's list of some class, each student on the list can be pre-loaded into Personal photo, each student name and student number.

S1002, each feature image is identified, obtains second feature number corresponding with the attendance object According to；

For example, carrying out recognition of face to each student's photo, the second feature data of corresponding student, i.e. face characteristic are obtained Data.

S1003, the title, second feature data and identity of each attendance object are stored in preset characteristic In library, subsequently into step S1004；

In general, second feature data, the title of identity and attendance object being stored in property data base, three There is one-to-one relationship, that is to say, that according to second feature data, that is, can determine corresponding identity and mesh between person Mark object oriented.

S1004, an acquisition at least frame wait for attendance image；

S1005, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S1006, each target object in the set of the target area is identified, is obtained and the target object pair The fisrt feature data answered；

S1007, from preset property data base, the determining second feature number to match with each fisrt feature data According to；

S1008, based on each second feature data to match, the identity documented by the property data base In, obtain target identities mark corresponding with the second feature data；

S1009, it is identified based on each target identities, is obtained and the target identities mark from the property data base Know corresponding target object title；

S1010, the target identities mark and target object name to each target object in attendance image that will acquire Claim, be determined as identity information corresponding with the target object, obtains with described to the corresponding identity information set of attendance image；

Here, it should be noted that step S1008 and step S1009 is actually the one of above-described embodiment step S905 Kind implements example.

S1011, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

In the embodiment of the present application, by obtained each second feature data to match, from the property data base institute In the identity of record, target identities mark corresponding with the second feature data is obtained；Based on each target body Part mark, obtains corresponding target object title from the property data base.Unique identification target pair is used in this way, first obtaining The target identities of elephant identify, and then, then identify the title for obtaining target object based on target identities, it is accurate that attendance can be improved Degree avoids the target object title obtained and the inconsistent problem of realistic objective object.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S1101 to step S1112:

S1101, title, feature image and the identity for being loaded into each attendance object recorded on the object set； Wherein, the identity is used for the corresponding attendance object of unique identification；

S1102, each feature image is identified, obtains second feature number corresponding with the attendance object According to；

S1103, second feature data and identity by each attendance object, are stored in the property data base；

S1104, using each identity as key assignments, by each key assignments and the corresponding attendance object oriented of the key assignments Hash table is written, the Hash table is stored in the property data base；

In order to improve attendance efficiency, the title of each attendance object and identity are usually pre-loaded into Hash table, that is, Using each identity as key assignments, Hash table is written into each key assignments and corresponding attendance object oriented.In subsequent S1109 In obtain target identities mark after, be identified as key assignments with each target identities, access the Hash table, obtain with it is described The corresponding target object title of key assignments.

Here, it should be noted that step S1103 and step S1104 is actually the one of above-described embodiment step S1003 Kind implements example.

S1105, an acquisition at least frame wait for attendance image；

S1106, adaptive partition processing is carried out to attendance image to described, obtains the target area for meeting predetermined identification condition Domain set；

S1107, each target object in the set of the target area is identified, is obtained and the target object pair The fisrt feature data answered；

S1108, from preset property data base, the determining second feature number to match with each fisrt feature data According to；

S1109, based on each second feature data to match, the identity documented by the property data base In, obtain target identities mark corresponding with the second feature data；

S1110, it is identified as key assignments with each target identities, accesses the Hash table, obtained and the target identities Identify corresponding target object title；

It is to be appreciated that being identified as key assignments after obtaining target identities mark with each target identities, accessing institute Hash table is stated, target object title corresponding with target identities mark is obtained, attendance efficiency can be improved.

S1111, the target identities mark and target object name to each target object in attendance image that will acquire Claim, be determined as identity information corresponding with the target object, obtains with described to the corresponding identity information set of attendance image；

S1112, based on preset object set and each to the corresponding identity information set of attendance image, generate attendance knot Fruit.

The embodiment of the present application provides another Work attendance method, the method comprising the steps of S1201 to step S1220:

S1201, the photo for being pre-loaded into the student number of each student and each student on object set；

S1202, recognition of face is carried out to the photo of each student, obtains second feature data corresponding with the student；

S1203, the second feature data and student number of each student are stored in preset property data base；

S1204, an acquisition at least frame wait for attendance image；

S1205, based on each face in the position on attendance image, determine the size of the face；

S1206, the size based on each face carry out subregion to attendance image to described, obtain set of image regions；

S1207, the size based on each face in described image region determine the flat of the face in described image region Equal size；

S1208, the average-size based on face in each described image region determine base from described image regional ensemble Quasi- image-region；

S1209, determine the benchmark image region face average-size, with face in each described image region Ratio between average-size；

S1210, it is based on the ratio, determines the amplification factor of image-region corresponding with the ratio；

S1211, the amplification factor based on each described image region, to image-region corresponding with the amplification factor into Row amplification obtains target area set；

S1212, each face in the set of the target area is identified, obtains the fisrt feature number of each face According to；

S1213, from preset property data base, the determining second feature number to match with each fisrt feature data According to；

S1214, it is obtained from student number documented by the property data base based on each second feature data to match Target student number corresponding with the second feature data is taken, is obtained to the corresponding target student number set of attendance image；

S1215, by the initial acquisition moment of attendance image to the duration between current time, be determined as attendance duration；

S1216, based on each to the corresponding target student number set of attendance image, determine and arrive class student number；

S1217, determine whether the attendance duration is less than preset duration；If so, executing step S1218；Otherwise, it executes Step S1219；

Whether it is less than preset number to class student number described in S1218, determination；If so, returning to step S1204； Otherwise, step S1219 is executed；

S1219, based on each to the corresponding target student number set of attendance image and preset object set, generate attendance knot Fruit；

If it is the classmate of Li Ming there are two in class, wherein there is a Li Ming not come to class, then, at this time if it is base Class student and student absent from school are determined in student name and object set, then are inaccurate, that is to say, that do not know on earth It is which Li Ming not next, so, it can be based on each target student number and preset object set here, generate attendance knot Fruit.

S1220, the checking-in result is sent to management platform.

In the embodiment of the present application, attendance image can be treated and carry out adaptive partition processing, made to every in attendance image One target object is all satisfied identification condition, to improve the recognition accuracy of target object, checking-in result is made to be more nearly reality Situation.The embodiment of the present application uses face recognition technology, provides a kind of check class attendance method that can once identify multiple faces, greatly The efficiency of attendance is improved greatly.And subregion algorithm is used, attendance can be completed during student normally attends class, it is real Real non-inductive attendance is showed.

Meanwhile to avoid using the Work attendance method based on server, bring improvement cost is high, and the later period needs special maintenance The problem of, the Work attendance method of the embodiment of the present application is realized by using intelligent fringe node, makes to implement the embodiment of the present application Work attendance method system become a lightweight system.

Therefore, the embodiment of the present application provides a kind of non-inductive classroom based on intelligent fringe node and face recognition technology and examines Diligent system meets the requirement in school instruction work to student attendance, while having accomplished the lightweight of system deployment, reduces the later period Maintenance cost.

A kind of Work attendance method based on intelligent fringe node and face identification system provided by the embodiments of the present application.This method The work attendance statistics work to student can be efficiently completed in the inductionless situation of student, also, checking-in result is sent back into clothes Business end is managed collectively.This method carries out terminal processes when implementing, through intelligent fringe node, reduces the transformation of equipment Expense and late maintenance cost make the attendance checking system for implementing this method accomplish that lightweight is disposed.

In addition, in order to improve system friendliness, the embodiment of the present application provides a kind of inductionless Work attendance method.It crosses at school Non-inductive recognition of face is carried out to student in journey, traditional attendance checking system is overcome and needs to carry out recognition of face to specified region Problem.

Also, the embodiment of the present application is based on a kind of adaptive partition processing method, will be collected according to camera placement information Image (i.e. to attendance image) be divided into several regions in front and back, interpolation magnification operation is carried out to the image in heel row region.Each region Image interpolation amplification factor carried out according to camera position and lens focus it is adaptive.This method can effectively overcome to attendance Front and rear row student face problem not of uniform size in image, to improve attendance accuracy rate.

The embodiment of the present application provides a kind of non-inductive check class attendance device, as shown in figure 5, the device 50 includes: that face is adopted Collect module 501, face recognition module 502 and attendance module 503；Wherein,

Face acquisition module 501 includes image acquisition units, image pre-processing unit and Face datection unit.

Image acquisition units: installing video camera immediately ahead of classroom, guarantees that the image of camera can cover entire classroom, with Just subsequent image procossing.Camera is linked on intelligent edge node devices, and is configured and is connected, so that intelligent edge Node device can receive the vision signal (i.e. described to attendance image) of video camera acquisition.

Image pre-processing unit: due in the scene of classroom, front and back face there is a problem of inconsistent, and therefore, it is necessary to rear It arranges image and carries out sub-regional interpolation operation.Its specific implementation is divided into two parts: first part determines number of partitions, according to classroom Classroom can be divided into 2 to 4 regions (i.e. described image region), then to each region by size and its front and back face difference Carry out different degrees of adaptive-interpolation amplification；Second part, when carrying out region division, for be likely to occur face segmentation Phenomenon (i.e. a face is divided in two regions), when dividing region, latter area can have with the picture material in previous region Overlapping to a certain extent, ensure that behind subregion in this way can also accurate detection to everyone.

As shown in fig. 6, classroom panorama 60 is divided for three regions, it is straight since final area is without any student information It connects and ignores the region.On heel row region 601 and front-seat region 602, in fact it can be found that second row of seats is in region division When covered；Meanwhile heel row region is amplified, and thus may insure that front and back region face size is similar substantially.

Face recognition module 502: when using the Work attendance device 50 for the first time, student's student number and its corresponding face figure are first loaded into Piece, which can automatically extract the feature of the face picture of loading, and be deposited in database, be permanentization guarantor It deposits.When carrying out recognition of face, the feature vector in the face feature vector and database currently extracted can be compared, Obtain confidence level；It is returned and the matched student number of face when confidence level is greater than certain threshold value.

Attendance module 503: during attendance, conventional method is that the student number that will be returned in identification process and database carry out Matching obtains student data.But the method efficiency is lower, needs continually to access database in identification process, reduces and examine Diligent speed.The present apparatus 50 uses Hash table, and the student number of student and name are loaded into Hash table in advance before attendance, identified Hash table need to be only accessed in the process, substantially increase the efficiency and speed of attendance.

Based on this, the embodiment of the present application provides the workflow of above-mentioned apparatus 50, as shown in fig. 7, the process includes:

After S71, attendance start, camera carries out Image Acquisition, obtains to attendance image；

S72, attendance image progress subregion adaptive-interpolation is treated, obtains target area set；

S73, Face datection is carried out to each target area in the set of target area, obtains corresponding face characteristic；

S74, the face characteristic for obtaining Face datection, in face database face carry out characteristic matching, obtain to The corresponding student's student number set of attendance image；

S75, determine whether attendance is completed；If so, executing step S76；Otherwise, S71 is returned to step；

It should be noted that if obtained in preset duration to class student number it is identical as preset number, attendance It completes, or if attendance duration is more than or equal to preset duration, attendance is completed, and executes step S76 at this time；If step S74 The number for student's student number that middle correspondence obtains is less than preset number, and attendance duration is less than preset duration, then attendance is not completed, S71 is returned to step at this time, is resurveyed an at least frame and is waited for attendance image.

S76, attendance terminate, and checking-in result generates.

In other embodiments, it for step S72, treats attendance image and carries out subregion adaptive-interpolation, obtain target Regional ensemble, that is, picture portion domain adaptive interpolation method, related description are as follows:

During image preprocessing, picture portion domain adaptive-interpolation will do it, according to the position of camera, classroom seat The size that distribution situation and its student's face are presented in attendance image carries out adaptive image segmentation and interpolation.Interpolation side Method uses bilinear interpolation method in the embodiment of the present application, naturally it is also possible to be changed to any other interpolation side according to demand Method.

For example, acquisition in attendance image, first row student, intermediate row student, face of last row student are average Size is respectively s1, s2, s3, adaptive-interpolation strategy provided by the embodiments of the present application, as shown in figure 8, including the following contents:

If meeting condition (s1/s3 < 2) and (s1/s2 < 2) simultaneously, forward and backward 2 figures will be divided into attendance image averaging As region, front-seat image-region is remained unchanged, 2 times of heel row image-region interpolation amplification.

If meeting condition (s1/s3<4) and (s1/s2>2) simultaneously, 3 before, during and after being divided into attendance image averaging A image-region, front-seat image-region remain unchanged, and the amplification factor of intermediate image area and heel row image-region is 2 times respectively With 3 times.

If meeting condition (s1/s3 > 4) and (s1/s2 > 2) simultaneously, before being divided into attendance image averaging, in front of, in Afterwards, 4 image-regions, front-seat image-region remain unchanged afterwards, in preceding image-region, in after image-region, heel row image-region Interpolation amplification multiple is 2 times, 3 times and 4 times respectively.

Under the conditions of other, without sub-regional interpolation processing.

In the embodiment of the present application, by carrying out plurality of human faces identification in intelligent fringe node, so that attendance work is completed, and Previous attendance is then carried out by the attendance mode that APP or individual human face identify；The Work attendance method of the embodiment of the present application supports nothing Incude attendance, does not need to carry out attendance by informing or specific mode, attendance work can be completed on classroom, will not be disturbed Disorderly normal teaching.And previous technology realization then needs that check class attendance can be carried out by taking pictures to specified region；In addition, Work attendance method provided by the embodiment of the present application is mainly treated attendance image by subregional mode and is handled, effective to solve Camera imaging of having determined is near big and far smaller to cause rear classmate's face too small, the case where can not being detected, to improve attendance Accuracy.

Work attendance method provided by the embodiments of the present application is supported to carry out face knowledge by the image of acquisition in classroom environment Not, so that the inductionless routine attendance check completed to student works.Due to solving by the way of adaptive partition processing The too small problem of heel row student's face, greatly improves the accuracy rate of attendance.Routine attendance check in teaching can use the application The Work attendance method of embodiment is substituted, and after the completion of attendance, checking-in result is carried out to unified management, and generate corresponding report Table can make teacher that the time of saving is used in teaching.

Based on embodiment above-mentioned, the embodiment of the present application provides a kind of Work attendance device, which includes included each mould Each unit included by block and each module can be realized by the processor in electronic equipment；It certainly can also be by specific Logic circuit realize；In the process of implementation, processor can be central processing unit (CPU), microprocessor (MPU), number Signal processor (DSP) or field programmable gate array (FPGA) etc..

Fig. 9 is the composed structure schematic diagram of the embodiment of the present application Work attendance device, as shown in figure 9, described device 900 includes:

Image capture module 901 is configured that an acquisition at least frame waits for attendance image；

Image processing module 902 is configured that and carries out adaptive partition processing to attendance image to described, obtains meeting pre- Surely the target area set of condition is identified；

Object Identification Module 903 is configured that and identifies to each target object in the set of the target area, obtains To described to the corresponding identity information set of attendance image；

Checking-in result generation module 904 is configured that based on preset object set and each to the corresponding body of attendance image Part information aggregate, generates checking-in result.

In other embodiments, described image processing module 902, comprising:

Multidomain treat-ment submodule is configured that and carries out subregion to attendance image to described, obtains set of image regions, described Image-region in set of image regions includes at least a target object；

Enhanced processing submodule is configured that and amplifies to the image-region in described image regional ensemble, met Gather the target area of predetermined identification condition.

In other embodiments, the multidomain treat-ment submodule includes:

Size determination unit is configured that based on each target object in the position on attendance image, determines The size of the target object；

Multidomain treat-ment unit is configured that the size based on each target object, divides to attendance image described Area obtains set of image regions.

In other embodiments, the enhanced processing submodule includes:

Amplification factor determination unit is configured that the size based on each target object in each described image region, really Fixed amplification factor corresponding with described image region；

Magnification processing is configured that the amplification factor based on each described image region, to the amplification factor pair The image-region answered amplifies, and obtains target area set.

In other embodiments, the checking-in result generation module 904 is configured that if the attendance duration is less than in advance If duration, and the target object number is equal to preset number, is based on preset object set and each to attendance image correspondence Identity information set, generate checking-in result；If the attendance duration is more than or equal to preset duration, it is based on preset object set It closes and each to the corresponding identity information set of attendance image, generation checking-in result.

In other embodiments, described device further include: sending module 905 is configured that and is sent to the checking-in result Manage platform.

In other embodiments, the Object Identification Module 903, comprising:

Object identifying submodule, be configured that the target area set in each target object identify, Obtain fisrt feature data corresponding with the target object；

Identity determines submodule, is configured that fisrt feature data and preset feature based on each target object Database determines identity information corresponding with the target object, obtains described to the corresponding identity information set of attendance image.

In other embodiments, the identity determines that submodule includes:

Characteristics determining unit is configured that from preset property data base, determination matches with each fisrt feature data Second feature data；

Identity determination unit is configured that based on each second feature data to match, obtains from the property data base Identity information corresponding with the second feature data is taken, is obtained described to the corresponding identity information set of attendance image.

In other embodiments, described device further includes that information insmods 904 and information storage module 905；Wherein,

The information insmods 904, is configured that the feature for being loaded into each attendance object recorded on the object set The title and identity of picture, each attendance object；Wherein, the identity is for attendance object described in unique identification；

The Object Identification Module 903 is configured that and identifies to each feature image, obtains and the attendance The corresponding second feature data of object；

The information storage module 905 is configured that the title, second feature data and identity mark of each attendance object Knowledge is stored in preset property data base.

In other embodiments, the identity determination unit, comprising:

Identity obtains subelement, each second feature data to match is configured to, from the characteristic In identity documented by library, target identities mark corresponding with the second feature data is obtained；

Object oriented obtains subelement, is configured that and is identified based on each target identities, from the property data base Obtain target object title corresponding with target identities mark；

Identity information determines subelement, is configured that the target to each target object in attendance image that will acquire Identity and target object title are determined as identity information corresponding with the target object, obtain with described to attendance figure As corresponding identity information set.

In other embodiments, the information storage module 905 is configured that the second feature data of each attendance object And identity, it is stored in the property data base；Using each identity as key assignments, by each key assignments and with it is described Hash table is written in the corresponding attendance object oriented of key assignments, and the Hash table is stored in the property data base.

In other embodiments, the target object is face.

The description of apparatus above embodiment, be with the description of above method embodiment it is similar, have same embodiment of the method Similar beneficial effect.For undisclosed technical detail in the application Installation practice, the application embodiment of the method is please referred to Description and understand.

It should be noted that in the embodiment of the present application, if realizing above-mentioned attendance side in the form of software function module Method, and when sold or used as an independent product, it also can store in a computer readable storage medium.Based on this The understanding of sample, the part that the technical solution of the embodiment of the present application substantially in other words contributes to the relevant technologies can be with software The form of product embodies, which is stored in a storage medium, including some instructions use so that One electronic equipment executes all or part of each embodiment the method for the application.And storage medium above-mentioned includes: U Disk, mobile hard disk, read-only memory (Read Only Memory, ROM), magnetic or disk etc. are various to can store program generation The medium of code.It is combined in this way, the embodiment of the present application is not limited to any specific hardware and software.

Accordingly, the embodiment of the present application provides a kind of electronic equipment, and Figure 10 is one kind of the embodiment of the present application electronic equipment Hardware entities schematic diagram, as shown in Figure 10, the hardware entities of the electronic equipment 100 include: including memory 1001 and processor 1002, the memory 1001 is stored with the computer program that can be run on processor 1002, and the processor 1002 executes Step in the Work attendance method provided in above-described embodiment is provided when described program.

Memory 1001 is configured to store the instruction and application that can be performed by processor 1002, can also cache device to be processed 1002 and electronic equipment 100 in each module it is to be processed or processed data (for example, image data, audio data, voice Communication data and video communication data), flash memory (FLASH) or random access storage device (Random can be passed through AccessMemory, RAM) it realizes.

Accordingly, the embodiment of the present application provides a kind of computer readable storage medium, is stored thereon with computer program, should The step in Work attendance method provided by the above embodiment is realized when computer program is executed by processor.

It need to be noted that: the description of medium stored above and apparatus embodiments, with retouching for above method embodiment It is similar for stating, and has with embodiment of the method similar beneficial effect.For in the application storage medium and apparatus embodiments not The technical detail of disclosure please refers to the description of the application embodiment of the method and understands.

It should be understood that " one embodiment " or " embodiment " that specification is mentioned in the whole text mean it is related with embodiment A particular feature, structure, or characteristic includes at least one embodiment of the application.Therefore, occur everywhere in the whole instruction " in one embodiment " or " in one embodiment " not necessarily refer to identical embodiment.In addition, these specific features, knot Structure or characteristic can combine in any suitable manner in one or more embodiments.It should be understood that in the various implementations of the application In example, magnitude of the sequence numbers of the above procedures are not meant that the order of the execution order, the execution sequence Ying Yiqi function of each process It can be determined with internal logic, the implementation process without coping with the embodiment of the present application constitutes any restriction.Above-mentioned the embodiment of the present application Serial number is for illustration only, does not represent the advantages or disadvantages of the embodiments.

It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row His property includes, so that the process, method, article or the device that include a series of elements not only include those elements, and And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do There is also other identical elements in the process, method of element, article or device.

In several embodiments provided herein, it should be understood that disclosed device and method can pass through it Its mode is realized.Apparatus embodiments described above are merely indicative, for example, the division of the unit, only A kind of logical function partition, there may be another division manner in actual implementation, such as: multiple units or components can combine, or It is desirably integrated into another system, or some features can be ignored or not executed.In addition, shown or discussed each composition portion Mutual coupling or direct-coupling or communication connection is divided to can be through some interfaces, the INDIRECT COUPLING of equipment or unit Or communication connection, it can be electrical, mechanical or other forms.

Above-mentioned unit as illustrated by the separation member, which can be or may not be, to be physically separated, aobvious as unit The component shown can be or may not be physical unit；Both it can be located in one place, and may be distributed over multiple network lists In member；Some or all of units can be selected to achieve the purpose of the solution of this embodiment according to the actual needs.

In addition, each functional unit in each embodiment of the application can be fully integrated in one processing unit, it can also To be each unit individually as a unit, can also be integrated in one unit with two or more units；It is above-mentioned Integrated unit both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.

Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through The relevant hardware of program instruction is completed, and program above-mentioned can store in computer-readable storage medium, which exists When execution, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: movable storage device, read-only deposits The various media that can store program code such as reservoir (ReadOnly Memory, ROM), magnetic or disk.

If alternatively, the above-mentioned integrated unit of the application is realized in the form of software function module and as independent product When selling or using, it also can store in a computer readable storage medium.Based on this understanding, the application is implemented The technical solution of example substantially in other words can be embodied in the form of software products the part that the relevant technologies contribute, The computer software product is stored in a storage medium, including some instructions are with so that an electronic equipment executes this Shen Please each embodiment the method all or part.And storage medium above-mentioned include: movable storage device, ROM, magnetic disk or The various media that can store program code such as person's CD.

The above, only presently filed embodiment, but the protection scope of the application is not limited thereto, it is any to be familiar with Those skilled in the art within the technical scope of the present application, can easily think of the change or the replacement, and should all cover Within the protection scope of the application.Therefore, the protection scope of the application should be based on the protection scope of the described claims.

Claims

1. a kind of Work attendance method, which is characterized in that the described method includes:

An acquisition at least frame waits for attendance image；

Adaptive partition processing is carried out to attendance image to described, obtains the target area set for meeting predetermined identification condition；

Each target object in the set of the target area is identified, is obtained with described to the corresponding identity of attendance image Information aggregate；

Based on preset object set and each to the corresponding identity information set of attendance image, checking-in result is generated.

2. the method according to claim 1, wherein described carry out at adaptive partition to described to attendance image Reason obtains the target area set for meeting predetermined identification condition, comprising:

Subregion is carried out to attendance image to described, obtains set of image regions, the image-region in described image regional ensemble is extremely It less include a target object；

Image-region in described image regional ensemble is amplified, the target area collection for meeting predetermined identification condition is obtained It closes.

3. according to the method described in claim 2, it is characterized in that, it is described to it is described to attendance image carry out subregion, obtain figure As regional ensemble, comprising:

According to it is described to each target object in attendance image in the position on attendance image, to described to attendance image Subregion is carried out, set of image regions is obtained.

4. according to the method described in claim 2, it is characterized in that, the image-region in described image regional ensemble into Row amplification obtains the target area set for meeting predetermined identification condition, comprising:

The size of the target object each in described image region is amplified to pre-set dimension, obtains target area set.

5. according to the method described in claim 3, it is characterized in that, it is described according to each target object in attendance image In the position on attendance image, subregion is carried out to attendance image to described, obtains set of image regions, comprising:

Based on each target object in the position on attendance image, the size of the target object is determined；

Based on the size of each target object, subregion is carried out to attendance image to described, obtains set of image regions.

6. according to the method described in claim 5, it is characterized in that, the image-region in described image regional ensemble into Row amplification obtains the target area set for meeting predetermined identification condition, comprising:

Based on the size of each target object in each described image region, times magnification corresponding with described image region is determined Number；

Based on the amplification factor in each described image region, image-region corresponding with the amplification factor is amplified, is obtained Gather to target area.

7. according to the method described in claim 6, it is characterized in that, each target based in each described image region The size of object determines amplification factor corresponding with described image region, comprising:

Based on the size of each target object in described image region, being averaged for the target object in described image region is determined Size；

Based on the average-size of the target object in described image region, benchmark image area is determined from described image regional ensemble Domain；

The average-size for determining the target object in the benchmark image region, the average ruler with target object in described image region Ratio between very little；Based on the ratio, the amplification factor in described image region is determined.

8. a kind of Work attendance device characterized by comprising

Image capture module is configured that an acquisition at least frame waits for attendance image；

Image processing module is configured that and carries out adaptive partition processing to attendance image to described, obtains meeting predetermined identification item Gather the target area of part；

Object Identification Module, be configured that the target area set in each target object identify, obtain with it is described To the corresponding identity information set of attendance image；

Checking-in result generation module is configured that based on preset object set and each to the corresponding identity information of attendance image Set generates checking-in result.

9. a kind of electronic equipment, including memory and processor, the memory are stored with the calculating that can be run on a processor Machine program, which is characterized in that the processor realizes any one of claim 1 to 7 Work attendance method when executing described program In step.

10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt The step in any one of claim 1 to 7 Work attendance method is realized when processor executes.