Real-time online comprehensive judgment method and system based on face recognition
By combining periodic and random image acquisition methods with power consumption and motion recognition thresholds, the problems of high power consumption and unreasonable judgment in video learning or work monitoring are solved, achieving low-power and high-efficiency monitoring results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JIANGXI SMALL VISION TECH CO LTD
- Filing Date
- 2018-12-14
- Publication Date
- 2026-05-22
AI Technical Summary
Existing technologies suffer from power waste and inaccurate judgments in video learning or work monitoring, especially on mobile devices. Continuous monitoring leads to high power consumption, and it is difficult to ensure monitoring effectiveness when leaving for short periods or during small-group discussions.
The method adopts a combination of periodic and random data collection, setting power consumption and frequency thresholds, and combining action recognition to ensure monitoring effectiveness and reduce power consumption. Random data collection increases the frequency of image acquisition to cope with acquisition failures, and action recognition sets a percentage threshold to ensure the rationality of judgments.
While ensuring monitoring effectiveness, reduce system power consumption, improve the rationality of judgment, avoid misjudgments, and achieve low-power and high-efficiency monitoring.
Smart Images
Figure CN109684963B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of facial recognition, and more specifically to a real-time online comprehensive judgment method and system based on facial recognition. Background Technology
[0002] Currently, there are many situations where individuals need to conduct video learning or work through client applications (such as PCs, mobile apps, etc.). For example, companies publish video courseware on corporate culture online, and employees usually learn on the client application. Other examples include driver's license training or other types of training.
[0003] The existing monitoring method involves continuously monitoring the face or fingerprint attendance of learners via a client application, which has two problems:
[0004] First, it wastes power (especially on mobile devices, where software and hardware need to be kept running continuously).
[0005] Secondly, relying solely on continuous facial recognition is insufficient to ensure real-time accuracy. For instance, it's unreasonable to deem short absences like pausing at lunchtime, answering phone calls, or going to the restroom as unacceptable. Similarly, in cases of short, small-group discussions involving two or more people, unclear facial recognition can also lead to disqualification, which is also unacceptable. Summary of the Invention
[0006] The purpose of this invention is to provide a real-time online comprehensive judgment method and system based on face recognition. It mainly uses periodic data collection, supplemented by random data collection. When data collection fails, the frequency of collecting target face images is increased. At the same time, power consumption thresholds and frequency thresholds are set to control the power consumption required for monitoring in real time, so as to minimize system power consumption while ensuring monitoring effectiveness. In addition, further action recognition is performed on the collected images, and a proportion threshold is set to ensure the rationality of the judgment.
[0007] To achieve the above objectives, in combination Figure 1 This invention proposes a real-time online comprehensive judgment method based on face recognition, the method comprising the following steps:
[0008] S1: Create a random data collection module and a periodic data collection module;
[0009] S2: Upon receiving a monitoring command, the periodic acquisition module is invoked to acquire target facial images and perform identity recognition according to a set period T1.
[0010] In response to failure to acquire target face images and / or failure of identity recognition, the random acquisition module is invoked. Based on periodic acquisition, a random acquisition variable is introduced to increase the frequency of target face image acquisition.
[0011] If the number of failures to acquire the target face image and / or the number of consecutive failures to identify the identity exceeds a preset threshold M, the target is determined to be lost; otherwise, the random acquisition module is stopped and periodic acquisition is resumed.
[0012] S3: Determine whether the target is qualified based on the collection results.
[0013] In a further embodiment, step S3, the method for determining whether the target is qualified based on the collection results, includes:
[0014] If the target is missing, the object is deemed unqualified. Otherwise, action recognition is performed on all face images of any object. If the proportion of face images whose action recognition does not meet the requirements exceeds a set threshold, the object is deemed unqualified.
[0015] In a further embodiment, the set percentage threshold is 20%.
[0016] In a further embodiment, the method further includes:
[0017] The entire online monitoring time is divided into N consecutive time intervals, and each time interval is assigned a corresponding weight factor ω. i i = 1, 2, ..., N
[0018] Count the number L of face images that do not meet the action recognition requirements within each time interval. i Objects that satisfy the following formula will be deemed unqualified:
[0019]
[0020] Where F is the total number of all face images of the object collected, and γ is a set percentage threshold.
[0021] In a further embodiment, the weighting factor satisfies the following condition: ω i+1 <ω i .
[0022] In a further embodiment, in step S2, the method of invoking the random acquisition module in response to the failure to acquire the target face image and / or the failure of identity recognition, and introducing a random acquisition variable based on periodic acquisition to increase the acquisition frequency of the target face image, includes:
[0023] While acquiring target face images according to a set period T1, the variable period T is also used. x Acquire target face image, variable period T x The following conditions must be met:
[0024]
[0025] Where x is the number of consecutive failures in acquiring the target face image and / or identity recognition, and k and a are the periodicity T of the variables. x The parameters are: P0 is the power consumption of a single acquisition of the target face image, and P is the preset power consumption threshold.
[0026] When P0×(x+1)≥P, stop random monitoring and determine the target object as unqualified.
[0027] In a further embodiment, k and a can be adjusted.
[0028] In a further embodiment, the failure to acquire the target face image includes no target to be identified or unclear target identification.
[0029] In a further embodiment, the identity recognition method includes:
[0030] The captured target face image is compared with the face image corresponding to the target login ID to identify the target's identity.
[0031] Based on the aforementioned real-time online comprehensive judgment method based on face recognition, this invention also mentions a real-time online comprehensive judgment system based on face recognition, characterized in that the system includes:
[0032] Random data collection module and periodic data collection module;
[0033] Module used to receive monitoring commands;
[0034] This module is used to retrieve periodic acquisition modules, acquire target face images according to a set period T1, and perform identity recognition.
[0035] This module is used to respond to failures in acquiring target face images and / or failures in identity recognition, and to invoke the random acquisition module. Based on periodic acquisition, it introduces random acquisition variables to increase the acquisition frequency of target face images.
[0036] Module used to determine if a target is lost;
[0037] This module is used to stop the random data collection module and revert to periodic data collection.
[0038] This module is used to determine whether a target is qualified based on the collected results.
[0039] Compared with existing technologies, the significant advantages of the above-described technical solution of the present invention are as follows:
[0040] 1) Primarily collect data periodically, supplemented by random collection. When collection fails, increase the frequency of collecting the target face image.
[0041] 2) It also sets power consumption thresholds and number of times thresholds to control the power consumption required for monitoring in real time, which can minimize system power consumption while ensuring monitoring effectiveness;
[0042] 3) In addition, further action recognition is performed on the acquired images, and a percentage threshold is set to ensure the rationality of the judgment.
[0043] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below may be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other. Furthermore, all combinations of the claimed subject matter are considered part of the inventive subject matter of this disclosure.
[0044] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description
[0045] The accompanying drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings, wherein:
[0046] Figure 1 This is a flowchart of the real-time online comprehensive judgment method based on face recognition of the present invention. Detailed Implementation
[0047] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.
[0048] Various aspects of the invention are described in this disclosure with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. The embodiments of this disclosure are not necessarily defined to include all aspects of the invention. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.
[0049] Combination Figure 1 This invention discloses a real-time online comprehensive judgment method based on face recognition, the method comprising the following steps:
[0050] S1: Create a random data collection module and a periodic data collection module.
[0051] S2: When a monitoring command is received, the periodic acquisition module is invoked to acquire the target face image and perform identity recognition according to a set period T1.
[0052] In response to failure to acquire target face images and / or failure of identity recognition, the random acquisition module is invoked. Based on periodic acquisition, a random acquisition variable is introduced to increase the frequency of target face image acquisition.
[0053] If the number of failed attempts to acquire the target face image and / or the number of consecutive failed identity recognition attempts exceeds a preset threshold M, the target is determined to be lost; otherwise, the random acquisition module is stopped and periodic acquisition is resumed.
[0054] In this embodiment, the failure to acquire the target face image includes no target to be identified, unclear target identification, such as insufficient image quality or face orientation not meeting the requirements, resulting in the inability to extract sufficient face feature values, etc.
[0055] The identity verification method includes:
[0056] The captured target facial image is compared with the facial image corresponding to the target's login ID to identify the target's identity. Specifically, the target provides their own facial image during registration, which serves as the standard facial image corresponding to the registration ID. When identity verification is required, the captured target facial image is compared with the standard facial image corresponding to the login ID used by the target. If the comparison is successful, the identity verification is passed; otherwise, the identity verification fails.
[0057] In a further embodiment, in step S2, the method of invoking the random acquisition module in response to the failure to acquire the target face image and / or the failure of identity recognition, and introducing a random acquisition variable based on periodic acquisition to increase the acquisition frequency of the target face image, includes:
[0058] While acquiring target face images according to a set period T1, the variable period T is also used. x Acquire target face image, variable period T x The following conditions must be met:
[0059]
[0060] Where x is the number of consecutive failures in acquiring the target face image and / or identity recognition, and k and a are the periodicity T of the variables. x The parameters are: P0 is the power consumption of a single acquisition of the target face image, and P is the preset power consumption threshold.
[0061] When a data acquisition failure occurs during the data collection process, random data acquisition is introduced to increase the data acquisition frequency. This takes into account both the characteristic of random acquisition supplementing periodic acquisition and the rigor of monitoring for anomalies. For example, the more consecutive data acquisition failures occur, the higher the monitoring intensity needs to be. Therefore, we can set T... x Less than T1, meaning that between two periodic collections, there will be at least one random collection, and the frequency of random collections is determined by parameters k and a.
[0062] Preferably, k and a can be adjusted in the background to change the variable period T. x Duration.
[0063] Specifically, setting a power consumption threshold can ensure that the power consumed by the monitoring meets the low power consumption requirements. When P0×(x+1)≥P, random monitoring is stopped, and the target object is judged as unqualified.
[0064] S3: Determine whether the target is qualified based on the collection results.
[0065] In a further embodiment, step S3, the method for determining whether the target is qualified based on the collection results, includes:
[0066] If the target is missing, the object is deemed unqualified. Otherwise, action recognition is performed on all face images of any object. If the proportion of face images whose action recognition does not meet the requirements exceeds a set threshold, the object is deemed unqualified.
[0067] In a further embodiment, the set percentage threshold is 20%.
[0068] For example, the purpose of action recognition is to determine whether a person is studying attentively. The process of making this judgment using the aforementioned method is as follows:
[0069] The system collects facial images of people within the area in front of the acquisition device, performs on-site facial recognition (comparing and recognizing the face corresponding to the logged-in ID), counts the number of acquisitions and the number of successful recognitions, and makes a comprehensive judgment based on the results. The purpose is to confirm whether more than 80% of the time is spent studying diligently.
[0070] In other examples, considering the impact of video playback duration on user energy and attention, we set the method as follows:
[0071] The entire online monitoring time is divided into N consecutive time intervals, and each time interval is assigned a corresponding weight factor ω. i i = 1, 2, ..., N
[0072] Count the number L of face images that do not meet the action recognition requirements within each time interval. iObjects that satisfy the following formula will be deemed unqualified:
[0073]
[0074] Where F is the total number of all face images of the object collected, and γ is a set percentage threshold.
[0075] For example, during the training process, as the training duration increases, the proportion of captured images showing trainees studying attentively decreases. Considering this, we set the weighting factor to satisfy the following condition: ω i+1 <ω i .
[0076] Based on the aforementioned real-time online comprehensive judgment method based on face recognition, this invention also mentions a real-time online comprehensive judgment system based on face recognition, the system comprising the following modules:
[0077] 1) Random data collection module and periodic data collection module.
[0078] 2) Module for receiving monitoring instructions.
[0079] 3) A module used to retrieve periodic acquisition modules, acquire target face images according to a set period T1, and perform identity recognition.
[0080] 4) A module used to respond to failure to acquire target face images and / or failure to identify the target face, and to invoke the random acquisition module to increase the acquisition frequency of target face images by introducing random acquisition variables based on periodic acquisition.
[0081] 5) Module used to determine if a target is lost.
[0082] 6) Used to stop the random data collection module and revert to the periodic data collection module.
[0083] 7) A module used to determine whether a target is qualified based on the collected results.
[0084] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.
Claims
1. A real-time online comprehensive judgment method based on face recognition, characterized in that, The method includes the following steps: S1: Create a random data collection module and a periodic data collection module; S2: Upon receiving a monitoring command, the periodic acquisition module is invoked to acquire target facial images and perform identity recognition according to a set period T1. In response to failure to acquire target face images and / or failure of identity recognition, the random acquisition module is invoked. Based on periodic acquisition, a random acquisition variable is introduced to increase the frequency of target face image acquisition. If the number of failures to acquire the target face image and / or the number of consecutive failures to identify the identity exceeds a preset threshold M, the target is determined to be lost; otherwise, the random acquisition module is stopped and periodic acquisition is resumed. S3: Determine whether the target is qualified based on the collection results. Specifically, this includes: determining that the target is unqualified if the target is missing; otherwise, perform action recognition on all face images of any object that has been collected. If the proportion of face images whose action recognition does not meet the requirements exceeds a set proportion threshold, then the object is determined to be unqualified. The method further includes: The entire online monitoring time is divided into N consecutive time intervals, and each time interval is assigned a corresponding weight factor ω. i i = 1, 2, ..., N Count the number L of face images that do not meet the action recognition requirements within each time interval. i Objects that satisfy the following formula will be deemed unqualified: Where F is the total number of all face images of the object collected, and γ is a set percentage threshold; In step S2, the method of responding to the failure to acquire the target face image and / or the failure of identity recognition by calling the random acquisition module and introducing random acquisition variables based on periodic acquisition to increase the acquisition frequency of the target face image includes: While acquiring target face images according to a set period T1, the variable period T is also used. x Acquire target face image, variable period T x The following conditions must be met: Where x is the number of consecutive failures in acquiring the target face image and / or identity recognition, and k and a are the periodicity T of the variables. x The influencing parameters are: P0 is the power consumption of a single acquisition of the target face image, and P is the preset power consumption threshold. When P0×(x+1)≥P, stop random monitoring and determine the target object as unqualified.
2. The real-time online comprehensive judgment method based on face recognition according to claim 1, characterized in that, The set percentage threshold is 20%.
3. The real-time online comprehensive judgment method based on face recognition according to claim 1, characterized in that, The weighting factor satisfies the following condition: oh i+1 <oh i 。 4. The real-time online comprehensive judgment method based on face recognition according to claim 1, characterized in that, The values of k and a are adjustable.
5. The real-time online comprehensive judgment method based on face recognition according to claim 1, characterized in that, The failure to acquire the target face image includes no target to be identified and unclear target identification.
6. The real-time online comprehensive judgment method based on face recognition according to claim 1, characterized in that, The identity verification method includes: The captured target face image is compared with the face image corresponding to the target login ID to identify the target's identity.
7. A real-time online comprehensive judgment system based on face recognition for implementing the real-time online comprehensive judgment method based on face recognition as described in claim 1, characterized in that, The system includes: Random data collection module and periodic data collection module; Module used to receive monitoring commands; This module is used to retrieve periodic acquisition modules, acquire target face images according to a set period T1, and perform identity recognition. This module is used to respond to failures in acquiring target face images and / or failures in identity recognition, and to invoke the random acquisition module. Based on periodic acquisition, it introduces random acquisition variables to increase the acquisition frequency of target face images. Module used to determine if a target is lost; This module is used to stop the random data collection module and revert to periodic data collection. This module is used to determine whether a target is qualified based on the collected results.