TWI621071B - Access control system for license plate and face recognition using deep learning - Google Patents

Abstract

The invention discloses an access control management system using deep learning for license plate and face recognition, which includes a vehicle access control device, an image capture module, an image feature database and an information processing device. Vehicle access control equipment is located at the gate of the driveway. The image capture module is used to capture the license plate image of the vehicle and the face image of the user. The image feature database establishes face feature data and license plate feature data; each face feature data and license plate feature data are set with identification data. The information processing device has a built-in image recognition module for image recognition processing of the license plate image and the face image, and identifies the matching facial feature data and license plate feature data in the image feature database to judge the license plate and the driver Whether it is a member of a community resident and the result of the judgment is yes, driving the lane gate is opened, in addition to effectively reducing the probability of theft of community vehicles, and improving the efficiency and security of vehicle access control.

Description

Access control system for license plate and face recognition using deep learning

The invention relates to an access control management system using deep learning for license plate and face recognition, and in particular to a community access management technology that can effectively reduce the probability of theft of community vehicles.

According to the general community (such as regional residential settlements, sky-high housing complexes, office buildings, apartment buildings, etc.), most of them will install driveway access control devices at the driveway gates of parking lots to identify people; or vehicles entering and leaving For regulatory purposes. At present, a more common method is to set a group of RFID card readers on the lane access control equipment. The residents who want to enter and exit the lane gates need only move the RFID card or RFID tag with them to the RFID card reader. Position, here, the identification unit built in the RFID card reader can then identify the password built in the RFID card or RFID tag. When the built-in password is a registered password, the identification unit will The fence drive is raised, so the occupant driver can enter or leave the driveway gate. Although this conventional technology has a lane access control function; however, when a resident accidentally loses an RFID card or an RFID tag, the password is profiled, so that a thief can easily use the profiled password or the lost RFID. Cards and RFID tags are used to duplicate RFID cards and RFID tags, so thieves can easily enter the parking lot of the community to steal vehicles; or property, resulting in poor access control of lanes.

In addition, in order to improve the security and confidentiality of lane access control, some minority communities have adopted voiceprint sensing, fingerprint sensing or retinal sensing to replace the above-mentioned RFID identification technology. Because the cost of voiceprint sensing, fingerprint sensing, or retinal sensing is too expensive, coupled with difficult installation and complicated processes, it has not been favored and adopted by related industries, resulting in a large commercial value. For lowering.

In order to improve the above-mentioned shortcomings, the relevant technical field has developed a patent as shown in the national invention No. I402777 "Community Building Real-time Monitoring and Management Method", which is stored in the real-life photo database and the real-life photos of the vehicle are stored in the community The building's vehicle entrance and exit are provided with a recognition sensor unit. After the management software is started, the information of the access control person is obtained from the identification data to the text database, and the real image database is used to obtain the real photo of the access control person and the real image photo of the vehicle, and it is displayed on the display screen connected to the monitoring computer A real image display interface and a text data display interface. Although the patent can display the real photo of the access control person, the real photo of the vehicle, and the personal text information of the access control person when the recognition sensor unit is triggered, it is ultimately the community that determines whether to open or control the access control of the vehicle. Guard; or administrator; In short, the patent is only used to reduce the identification burden of community guards or administrators on the access control of people or vehicles. It can be seen that the patent does not automatically determine by image recognition. Whether to open or control vehicle access control entrances. Once community guards or administrators are stunned, careless, or lazy to fall asleep, there may be dead ends of vehicle access control. As a result, thieves may invade community parking lots and steal vehicles. As the situation arose, it can be seen that the patent is indeed not yet perfect, and there is still a need for further improvement.

So far, no patent or paper has been proposed for access control systems that use deep learning for license plate and face recognition, and based on the urgent needs of related industries Next, the inventor has developed a set of the present invention that is different from the above-mentioned conventional technologies and patents through continuous efforts in research and development.

The first object of the present invention is to provide an access control management system using a license plate and face recognition, which is mainly set by functions such as license plate and face recognition. In addition to effectively reducing the risk of theft of community vehicles, it can also Improve vehicle access control efficiency and safety. The technical means adopted to achieve the first object of the invention include vehicle access control equipment, image capture modules, image feature databases and information processing devices. Vehicle access control equipment is located at the gate of the driveway. The image capture module is used to capture the license plate image of the vehicle and the face image of the user. The image feature database is established to include a plurality of facial feature data and license plate feature data; each facial feature data and license plate feature data are set with identification data. The information processing device has a built-in image recognition module for image recognition processing of the license plate image and the face image, and identifies the matching facial feature data and license plate feature data in the image feature database to judge the license plate and the driver Whether it is a member of a community resident and the result of the judgment is yes, then the driveway gate is opened.

A second object of the present invention is to provide an access control management system capable of restricting a stolen vehicle to a lane gate, mainly limiting a stolen vehicle in a community to a lane gate, so as to avoid reversing the vehicle due to the stolen vehicle in the process of hunting a thief. To the occurrence of vehicle collision damage caused by the parking lot, thereby effectively reducing the loss of community property. The technical means adopted to achieve the second object of the invention include vehicle access control equipment, image capture module, image feature database and information processing device. Vehicle access control equipment is located at the gate of the driveway. The image capture module is used to capture the license plate image of the vehicle and the face image of the user. The image feature database is established to include multiple facial feature data and license plate feature data; each facial feature data and license plate feature data are set to have an identification data. The information processing device has a built-in image recognition module for image recognition processing of the license plate image and the face image, and identifies the matching facial feature data and license plate feature data in the image feature database to judge the license plate and the driver Whether it is a member of a community resident and the result of the judgment is yes, then the driveway gate is opened. Wherein, the vehicle access control device further includes a blocking mechanism provided at the gate of the lane; the vehicle access control device is provided with an alarm device electrically connected to the information processing device; when the user inputs the password / body characteristics When the password or the human body feature entered by the device is incorrect, the information processing device drives the alarm device to issue an alarm signal, and drives the blocking mechanism to block the lane gate.

A third object of the present invention is to provide an access control management system that can automatically adjust the image capture position according to the vehicle style to increase the probability of successful image recognition. The technical means adopted to achieve the third purpose of the invention include vehicle access control equipment, image capture modules, image feature databases and information processing devices. Vehicle access control equipment is located at the gate of the driveway. The image capture module is used to capture the license plate image of the vehicle and the face image of the user. The image feature database is established to include a plurality of facial feature data and license plate feature data; each facial feature data and license plate feature data are set with identification data. The information processing device has a built-in image recognition module for image recognition processing of the license plate image and the face image, and identifies the matching facial feature data and license plate feature data in the image feature database to judge the license plate and the driver Whether it is a member of a community resident and the result of the judgment is yes, then the driveway gate is opened. Wherein, each of the license plate characteristic data is set with a displacement travel data; the vehicle access control device further includes a mount, a displacement driving mechanism and a position identification module; the displacement driving mechanism is used to drive the carriage to Reciprocating displacement in at least one direction; the position recognition module is used to sense whether the vehicle is positioned at a predetermined position; as a result, a trigger signal is generated to the information processing device, and the information processing device is Read the displacement stroke data set by the license plate characteristic data of the identified license plate, and output a corresponding displacement stroke driving signal to the displacement driving mechanism to drive the carrier and the second image capture module After positioning to the predetermined position, after the vehicle leaves the lane gate, the information processing device drives the displacement driving mechanism to return the carrier and the second image capturing module to the original position.

The fourth object of the present invention is to provide an access control management system using deep learning for license plate and face recognition, mainly through the function settings of deep learning, license plate and face recognition, in addition to effectively reducing the risk of theft of community vehicles. In addition to the probability, it can improve the efficiency and safety of vehicle access control. The technical means adopted to achieve the fourth objective of the invention includes vehicle access control equipment, image capture module, image feature database and information processing device. Vehicle access control equipment is located at the gate of the driveway. The image capture module is used to capture the license plate image of the vehicle and the face image of the user. The image feature database is established to include a plurality of facial feature data and license plate feature data; each facial feature data and license plate feature data are set with identification data. The information processing device has a built-in image recognition module for image recognition processing of the license plate image and the face image, and identifies the matching facial feature data and license plate feature data in the image feature database to judge the license plate and the driver Whether it is a member of a community resident and the result of the judgment is yes, then the driveway gate is opened. The image recognition module is a deep learning algorithm with a deep learning training function to perform image recognition processing. Executing the deep learning algorithm includes the following steps: a training phase step, which establishes a deep learning model, and Input the facial feature data, the license plate feature data, and a large number of the license plate feature images and the facial feature images into the deep learning model, and the deep learning model tests the correctness of image recognition, and then determines whether the correctness of image recognition Enough, when the judgment result is yes, the recognition result is output and stored; when the judgment result is no, the depth is made Degree learning model self-correcting learning; and a step of running a prediction phase, in which a deep learning model inputs the facial feature data, the license plate feature data, and the license plate feature image and the facial feature image captured in real time in the deep learning model, Predictive image recognition is performed by the deep learning model to obtain at least one recognition result that can identify whether the license plate and the driver are residents of the community.

1‧‧‧ vehicle

10‧‧‧lane gate

20‧‧‧vehicle access control equipment

21‧‧‧ Fence device

210‧‧‧ Fence

22‧‧‧stop mechanism

220‧‧‧ base

220a‧‧‧ Arrive

220b‧‧‧inclined plate

220c‧‧‧Electric tank room

221‧‧‧ Motor

221a‧‧‧Drive Gear

222‧‧‧Driven Gear

223‧‧‧ Bezel

224‧‧‧Drive shaft

23‧‧‧Displacement drive mechanism

230‧‧‧first displacement mechanism

231‧‧‧Second displacement mechanism

24‧‧‧Position recognition module

30‧‧‧Image capture module

40‧‧‧Image Feature Database

50‧‧‧ Information Processing Device

51‧‧‧Image recognition module

510‧‧‧Deep Learning Model

60‧‧‧Smart electronic device

61‧‧‧display

62‧‧‧Signal transmission means

70‧‧‧Password / human characteristics input device

71‧‧‧ Voice output device

72‧‧‧Alarm device

FIG. 1 is a functional block diagram of a first embodiment of the present invention.

FIG. 2 is a functional block diagram of a second embodiment of the present invention.

FIG. 3 is a functional block diagram of a third embodiment of the present invention.

FIG. 4 is a schematic diagram of implementation of a training phase of deep learning according to a fourth embodiment of the present invention.

FIG. 5 is a schematic diagram of implementation of a running prediction stage of deep learning according to a fourth embodiment of the present invention.

FIG. 6 is a screen display diagram of a lane management input interface of the present invention.

FIG. 7 is a schematic diagram of the operation of the blocking mechanism of the present invention.

FIG. 8 is a schematic diagram of another operation of the blocking mechanism of the present invention.

FIG. 9 is a schematic diagram of the top view of the blocking mechanism of the present invention.

In order to allow your reviewers to further understand the overall technical features of the present invention and achieve the effect of achieving the purpose of the present invention, specific embodiments and drawings are described in detail below: Please refer to Figures 1 and 7 to achieve A first embodiment of the first object of the present invention includes a vehicle access control device 20, an image capture module 30, and image feature information. Technical features such as the magazine 40 and the information processing device 50. The vehicle access control device 20 is provided on the lane gate 10 of a parking lot in a community, and is used to control the access of the vehicle 1 to the lane gate 10. The image capture module 30 is disposed on the vehicle access control device 20 to capture the license plate image of the vehicle 1 parked at the lane gate 10 and the face image of the user. The image feature database 40 is constructed to include a plurality of facial feature data and license plate feature data; each facial feature data and each license plate feature data is set with identification data representing one member of a community resident. The information processing device 50 has an image recognition module 51 built therein for image recognition processing of the license plate image and the face image, and recognizes facial feature data and facial image data that match the face image and the license plate image in the image feature database 40. License plate characteristic data to determine whether the license plate and driver of vehicle 1 at lane gate 10 are members of the residents of the community. When the judgment result is yes, the vehicle access control device 20 is driven to open lane gate 10, so this Members of the community resident can drive into or out of the lane gate 10; otherwise, when the judgment result is no, the lane gate 10 is closed, so the administrator can move forward to check further, if it is determined that they are not members of the community resident The vehicle 1 is required to leave the lane gate 10; or the suspicious vehicle 1 is confined in the lane gate 10 through the vehicle access control device 20, in order to further clarify the true identity of the vehicle 1 driving.

In an embodiment shown in FIGS. 2 and 6, the present invention further includes a smart electronic device 60 (such as a smart phone) for each member of the community, and a lane management is built in the smart electronic device 60. Software, when this lane management software is executed, a lane management input interface 610 is displayed on the display screen 61 of the smart electronic device 60, and this lane management input interface displays a resident member input interface, member photo upload interface, address input interface, License plate input interface, vehicle type input interface, and password key input interface; when the management data setting of the lane management input interface is completed, the smart electronic device 60 uses a signal transmission method 62 (such as mobile communication network) (Combination of road and mobile communication interface) transmits management data to the database of the information processing device 50 as a basis for image recognition processing.

Please refer to FIG. 2 for a second embodiment for achieving the second object of the present invention. In addition to the technical features of the first embodiment, the number of the above-mentioned image capturing modules 30 is two in this embodiment. An image capture module 30 is electrically connected to the information processing device 50 and is disposed on a first position (such as a fence) of the vehicle access control device 20 to capture a license plate image of the vehicle 1. The second image capture module 30 is electrically connected to the information processing device 50 and is disposed at a second position (as shown in FIG. 7) of the vehicle access control device 20 to capture a user's face image; The embodiment is mainly that the vehicle access control device 20 includes a fence device 21 that can be rotated and lifted. The fence device 21 includes a fence 210. The first position is located near the middle of the front of the fence 210 facing the vehicle 1. 20 is provided with a password / body characteristic input device 70 (such as an identification device such as an iris, a face, or a fingerprint) and a voice output device 71 (such as a combination of an audio processing circuit and a speaker) electrically connected to the information processing device 50, respectively. When the above determination result is negative, the information processing device 50 sends a voice signal requesting a password through the voice output device 71. When the password or the human body characteristic input by the user in the password / body characteristic input device 70 is correct, The information processing device 50 drives the fence 210 of the fence device 21 to rise, so that the lane gate 10 can be opened.

In addition, the vehicle access control device 20 shown in FIGS. 6 to 9 further includes a blocking mechanism 22 provided at a middle position of the lane gate 10 and an alarm device 72 electrically connected to the information processing device 50. When the password entered by the user in the password / body characteristic input device 70 or the body characteristic is wrong, the information processing device 50 drives the alarm device 72 to issue an alarm signal, and drives the blocking mechanism 22 to restrict the vehicle 1 to the lane gate. Within 10, here, you can avoid chasing During the process of thief arrest, the vehicle 1 was damaged by the collision of the stolen vehicle 1 and returned to the parking lot.

Specifically, the blocking mechanism 22 includes a base 220 extending in length; two motors 221, two driven gears 222 meshing with the driving gear 221a of the two motors 221, and a baffle 223 extending in length. The base 220 includes an abutment plate 220a, an inclined plate 220b juxtaposed at the rear end of the abutment plate 220a, and two electrical tank chambers 220c for housing two motors 221 and two driven gears 222. Two side walls of the baffle plate 223 protrude from two driving shafts 224, and the two driving shafts 224 pass through the electrical tank chamber 220c and are linked with the two driven gears 222. When the vehicle 1 is to be restricted to the lane gate 10, the information processing device 50 drives the second motor 221 to rotate in one direction (that is, counterclockwise), and drives the second driven gear 222 to drive the shutter 223 to the other direction (that is, clockwise). (Clockwise direction) rotation, when the baffle plate 223 abuts against the front end surface of the swash plate 220b (that is, the longitudinally extending end surface), a locked state is formed (even if the baffle plate cannot continue to rotate in another direction), here, That is, the front wheel of the vehicle 1 can be blocked by the raised baffle 223, so that the vehicle 1 cannot drive backward to leave the lane gate 10.

Please refer to FIG. 3 for a third embodiment for achieving the third purpose of the present invention. In addition to the technical features of the first embodiment, this embodiment includes preset displacement data for each member ’s identification data. (If you select a car, RV, or sports car on the car type input interface of the car type input interface, you can generate three different displacement travel data according to the selected car type). In addition, the vehicle access control device 20 further includes a carrier (not shown in the figure) for the second image capturing module 30, a displacement driving mechanism 23, and a position recognition module 24. The displacement driving mechanism 23 is used to drive the carrier to reciprocate in at least one direction. The position recognition module 24 (such as a reflective photo sensor; or a combination of a magnetic reed switch and a magnetic sheet) is used to sense whether the vehicle 1 is positioned at a predetermined position; If yes, a trigger signal is generated to the information processing device 50, and the information processing device 50 reads the displacement stroke data set by the license plate characteristic data of the recognized license plate, and outputs a corresponding displacement stroke driving signal to the displacement driving mechanism 23 In this case, the driving carrier and its two image capturing modules 30 are positioned to the predetermined position. Here, the second image capturing module 30 can be used to capture the facial image of the vehicle 1 driving. After the vehicle 1 leaves the lane gate 10, the information processing device 50 then drives the displacement driving mechanism 23 to return the carrier and its two image capturing modules 30 to their original positions.

More specifically, in order to achieve a more accurate image capture position, as shown in FIG. 3, the displacement driving mechanism 23 includes a front and rear support for driving the carriage and its two image capture modules 30 relative to the vehicle 1. A first displacement mechanism 230 moving in the direction and a second displacement mechanism 231 for driving the displacement driving mechanism 23 to move up and down relative to the vehicle 1. Here, the second image capturing module 30 can be used to capture the driving of the vehicle 1. Facial image.

Please refer to FIG. 1 and FIGS. 4 and 5 for a fourth embodiment for achieving the fourth object of the present invention. In addition to the technical features of the first embodiment, the image recognition module 51 described above is a type A deep learning algorithm with a deep learning training function to perform image recognition processing. The execution of a deep learning algorithm includes the following steps:

(a) The steps in the training phase are to establish a deep learning model 510, and input facial feature data, license plate feature data, and a large number of license plate images and facial images into the deep learning model 510. The deep learning model 510 tests the image recognition The accuracy rate is then judged whether the image recognition accuracy rate is sufficient. When the determination result is yes, the identification result is output and stored; when the determination result is no, the deep learning model 510 is caused to perform self-correction learning.

(b) Run the prediction phase step. The deep learning model 510 inputs facial feature data, license plate feature data, and real-time captured license plate images and facial images, and is performed by the deep learning model 510. Predictive image recognition to obtain at least one identification result that can identify whether the license plate and the driver are residents of the community.

The present invention analyzes the feature values of the feature image after image pre-processing by means of image recognition processing, and then compares the feature values in the image feature database 40 that are roughly consistent with the feature values (for example, the similarity of discrimination is about 70% The above) feature data and corresponding semantic data are obtained from the matched feature data; if the image recognition success rate is not high, the similarity of recognition can be increased until the required image recognition success rate is reached.

Preferably, the above-mentioned deep learning algorithm may be a convolutional neural network. After obtaining a feature image from an image capture device, it undergoes image preprocessing (that is, preprocessing), feature extraction, feature selection, and then inference and Make predictive identification. On the other hand, the essence of deep learning for convolutional neural networks is to build more machine learning models with multiple hidden layers and massive training data to achieve more useful features of learning, and ultimately improve the accuracy of classification or prediction. Convolutional neural networks use massive training data to learn feature recognition, where they can characterize the rich internal information of the data. Since the convolutional neural network is a network structure with shared weights, in addition to reducing the complexity of the network model, the number of weights can also be reduced. This advantage is more obvious when the input of the network is a multi-dimensional image, so that the image can be directly used as the input of the network, avoiding the complex feature extraction and data reconstruction process in traditional image recognition algorithms. Object classification methods are almost based on statistical features, which means that certain features must be extracted before discrimination. However, explicit feature extraction is not easy, and it is not always reliable in some application problems. Convolutional neural networks can avoid explicit feature sampling and implicitly learn from training data. This makes the convolutional neural network distinct from other neural network-based classifiers, and integrates feature extraction into multi-layer perceptrons by restructuring and reducing weights. It can process grayscale images directly, Can be used directly for image-based classification. Convolutional neural networks have the following advantages over general neural networks in image processing: the input image matches the topology of the network well; feature extraction and pattern classification are performed at the same time and are generated during training; Weight sharing can reduce the training parameters of the network, making the neural network structure simpler and more adaptable.

Further, the face recognition of the present invention may be a facial recognition technology based on facial features recognition. First, the facial image is cut into areas of facial features, including eyebrows, left eye, right eye, nose, mouth and Area images such as ears; then input these facial features regional images into pre-trained individual classifiers, and follow the preset threshold to determine which image category the regional images entered belong to, and finally these facial features The candidates identified in the regional image are combined, and then a vote is used to determine which community household member in the image feature database 40 the input facial image belongs to.

The above description is only a feasible embodiment of the present invention, and is not intended to limit the patent scope of the present invention. Any equivalent implementation of other changes based on the content, characteristics and spirit of the following claims should be It is included in the patent scope of the present invention. The structural features specifically defined in the present invention are not found in similar items, and are practical and progressive. They have met the requirements for invention patents. They have filed applications in accordance with the law. I would like to request the Bureau to verify the patents in accordance with the law in order to maintain this document. Applicants' legitimate rights and interests.

Claims (9)

An access control management system using deep learning for license plate and face recognition, which includes: at least one vehicle access control device, which is set on a lane gate to control vehicles' entry and exit at the lane gate; at least one image capture module , Which is set on the vehicle access control device to capture the license plate image of the vehicle parked at the lane gate and a user's face image; an image feature database, which contains a plurality of facial features Data and license plate feature data; each of the face feature data and each license plate feature data is set with identification data representing one of the members of the community household; and an information processing device, which has an image recognition module built in, Image recognition processing is performed on the license plate image and the face image, and the facial feature data and the license plate feature data that are consistent with the facial image and the license plate image are identified in the image feature database to determine whether the The license plate of the vehicle at the gate of the driveway and whether the driver is a member of the residents of the community. If the judgment is yes, the vehicle is driven to enter the gate Control device opens the lane gate; the number of the image capture modules is two; one of the image capture modules is electrically connected to the information processing device and is set at the first position of the vehicle access control device For capturing the license plate image of the vehicle; second, the image capture module is electrically connected to the information processing device and is set at a second position of the vehicle access control device to capture the user ’s Facial image; the vehicle access control device includes a fence device that can be rotated and raised, the fence device includes a fence, the first position is located near the middle of the front of the fence toward the vehicle; the vehicle access control device further includes a device A stop mechanism at the gate of the lane, the stop mechanism includes a base with an extended length, two motors, two driven gears that mesh with the driving gears of the two motors, and a length-extended baffle; the base package It includes an abutment plate, a sloping plate juxtaposed at the rear end of the abutment plate, and two electrical tanks for accommodating the two motors and the two driven gears; two side walls of the baffle protrude from two driving shafts, and the two driving shafts Passes through the electrical tank and is linked with the two driven gears; when the vehicle is to be prohibited from entering the lane gate, the information processing device drives the two motors to rotate in one direction, and drives the two driven gears in conjunction with the two driven gears The plate rotates in the other direction, and when the baffle abuts against the front surface of the swash plate, it becomes a locked state, and the front wheel of the vehicle can be locked by erecting the baffle so that the vehicle cannot be disabled. Leave the lane gate in reverse. The access control management system using deep learning for license plate and face recognition as described in claim 1, wherein a password / body feature input device and a voice that are electrically connected to the information processing device are provided in the vehicle access control device, respectively. Output device; when the judgment result is no, the information processing device sends a voice signal requesting a password through the voice output device, and when the password or the human body characteristic input by the user in the password / body characteristic input device is correct At that time, the information processing device drives the fence to rise to open the lane gate. The access control management system using deep learning for license plate and face recognition as described in claim 2, wherein the vehicle access control device is provided with an alarm device electrically connected to the information processing device; when the user enters the password / When the password input by the human body input device or the human body is wrong, the information processing device drives the alarm device to issue an alarm signal, and drives the blocking mechanism to inhibit the lane gate. The access control management system using deep learning for license plate and face recognition as described in claim 3, wherein the human characteristics are selected from one of iris, face and fingerprint. The access control management system using deep learning for license plate and face recognition as described in claim 1, wherein the identification data of each member includes a displacement stroke data set in advance The number of the image capture module is two; one of the image capture module is set at the first position of the vehicle access control device; the other is the image capture module is set at the vehicle access control device The second position of the vehicle; the vehicle access control device further includes a carrier for setting the second image capturing module, a displacement drive mechanism and a position identification module; the displacement drive mechanism is used to drive the carriage to A reciprocating displacement in at least one direction; the position recognition module is used to sense whether the vehicle is positioned at a predetermined position; when the determination result is yes, a trigger signal is generated to the information processing device, and the information processing device reads Take the displacement stroke data set by the license plate characteristic data that has identified the license plate, and output a corresponding displacement stroke driving signal to the displacement driving mechanism to drive the carrier and the two image acquisition module positioning At the predetermined position, when the vehicle leaves the lane gate, the information processing device drives the displacement driving mechanism to return the carrier and the second image capturing module to the original position. Position. The access control management system using deep learning for license plate and face recognition as described in claim 5, wherein the displacement driving mechanism includes a front drive for driving the carrier and two image capture modules for the vehicle, A first displacement mechanism moving in the backward direction, and a second displacement mechanism for driving the displacement driving mechanism to move up and down relative to the vehicle. As described in claim 1, the access control management system using deep learning for license plate and face recognition further includes a smart electronic device for each member of the household, and the smart electronic device has a lane management software built in, When the lane management software is executed, a lane management input interface is displayed on a display screen of the smart electronic device. The lane management input interface displays a selection from an address input interface, a household member input interface, a member photo upload interface, and a license plate input. At least three of the interface, the type of vehicle input interface, and the password key input interface; when the management data setting of the lane management input interface is completed, the smart electronic device transmits a signal via a signal The segment transmits the management data to the database of the information processing module as a basis for image recognition processing. The access control management system using deep learning for license plate and face recognition as described in claim 1, wherein the image recognition module is a deep learning algorithm with deep learning training function to perform image recognition processing, and execute the depth The learning algorithm includes the following steps: a training step, which establishes a deep learning model, and inputs the facial feature data, the license plate feature data, and a large number of the license plate image and the face image into the deep learning model, and The deep learning model tests the correctness of image recognition, and then determines whether the correct rate of image recognition is sufficient. When the judgment result is yes, the recognition result is output and stored; when the judgment result is no, the deep learning model is self-corrected. Learning; and a step of running a prediction phase, which involves inputting the facial feature data, the license plate feature data, and the license plate image and the facial image captured in real time in the deep learning model, and predictive images are performed by the deep learning model Identification to get at least one that can identify whether the license plate and the driver are in the community Identification of household members. An access control method using deep learning for license plate and face recognition, comprising: setting at least one vehicle access control device at at least one lane gate of a community to control vehicles entering and exiting the lane gate; capturing at least one image The module is set on the vehicle access control device to capture the license plate image of the vehicle parked at the gate of the lane and a user's face image; a feature database containing a plurality of facial features is established in an image feature database And license plate feature data; each of the facial feature data and each of the license plate feature data is provided with identification data representing one of the members of the community household; and An image processing module is built in an information processing device to perform image recognition processing on the license plate image and the face image, and identify the face image and the license plate image that match the face image and the license plate image in the image feature database. Facial feature data and license plate feature data to determine whether the license plate and driver of the vehicle at the lane gate are one of the members of the community resident. If the judgment is yes, then drive the vehicle access control device to open the lane gate. Among them, the number of the image capturing module is two; one of the image capturing module is electrically connected to the information processing device and is set at the first position of the vehicle access control device for capturing the The license plate image of the vehicle; second, the image capture module is electrically connected to the information processing device and is set at the second position of the vehicle access control device to capture the face image of the user; the vehicle The access control device includes a fence device capable of rotating and lifting, the fence device includes a fence, and the first position is located near the middle of the front of the fence facing the vehicle; the vehicle The prohibition control device further includes a stop mechanism provided at the gate of the lane, the stop mechanism includes a base extending in length, two motors, two driven gears meshing with the driving gears of the two motors, and a length extending stop Plate; the base includes an abutting plate, a sloping plate juxtaposed at the rear end of the abutting plate, and two electrical tanks for housing the two motors and the two driven gears; two side walls of the baffle protrude with two driving shafts The two drive shafts pass through the electrical slot and are linked with the two driven gears. When the vehicle is to be restrained from the lane gate, the information processing device drives the two motors to rotate in one direction and links the two slaves. The moving gear drives the baffle to rotate in the other direction. When the baffle abuts against the front surface of the swash plate, it becomes a locked state. The front wheel of the vehicle can be locked by standing up the baffle. This prevents the vehicle from reversing away from the lane gate.