A method and system for target re-identification suitable for night

Abstract

The application provides a method suitable for target re-identification at night, which comprises the following steps: training a standardization flow module; further pre-processing by the standardization flow module to generate the low-light training set X l ; collecting and obtaining a training set, which comprises an original training set X and a low-light training set X l ; and performing feature recognition on picture sample data in the training set by a reid network. The technical scheme of the application is directed to the adverse interference of night picture collection, a standardization flow module is introduced outside the normal recognition network to enhance the original collected pictures under low light, the recognition ability of the re-identification network for the low-light pictures is effectively improved by the module, and an effective technical scheme for realizing target re-identification under night conditions is provided.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer vision, and in particular relates to a method and system for target re-identification suitable for nighttime. Background Art

[0002] With the advancement of society and technology, facial recognition is becoming an increasingly reliable security technology. However, the resolution of most current security cameras often falls short of the requirements of facial recognition systems. Therefore, pedestrian re-identification technology that can be applied to existing surveillance systems is extremely necessary.

[0003] Person re-identification uses image processing technology to determine whether a pedestrian captured by one camera appears in other cameras, thereby depicting the pedestrian's activity path and achieving cross-camera tracking.

[0004] However, under low-light (nighttime) conditions, images captured by surveillance cameras often suffer from color shift, low contrast, noise interference, detail loss, and other problems. This image quality significantly reduces the recognition performance of existing target re-identification methods.

[0005] In view of this, it is very meaningful to propose a method and system for target re-identification suitable for nighttime. Summary of the Invention

[0006] In order to solve the problem that existing image collection in low-light (nighttime) environments causes huge interference to nighttime target re-identification, the present invention provides a method and system for nighttime target re-identification to solve the above-mentioned technical defects.

[0007] In a first aspect, the present invention provides a method for target re-identification at night, the method comprising the following steps:

[0008] Train the normalized flow module;

[0009] The low-light training set X is further generated by preprocessing through the standardized flow module l ;

[0010] Collect and obtain a training set, which includes an original training set X and a low-light training set X l ;

[0011] The image sample data in the training set is subjected to feature recognition through the ReID network.

[0012] Preferably, training the standardized flow module includes:

[0013] The normalized flow module is based on the low-light training set X l The low-light samples in are used as output to obtain an effective mapping from low-light to normal exposure images.

[0014] More preferably, it also includes:

[0015] During the first N epochs of training, the backbone network parameters are frozen, and only the normalized stream is trained first;

[0016] When the current epoch>N, the learning rate of the normalized flow module is adjusted to 1 / 10 of the backbone network.

[0017] Further preferably, by the sample pair (X l ,X) to measure the standard flow Θ, the loss function of the standard flow module is as follows:

[0018] L(X l ,X)=-logf flow (X l |X)

[0019] Among them, f flow (X l |X) represents the conditional probability density function of X.

[0020] Preferably, the low-light training set X l The original training set X is generated by adjusting the pixel values ​​of each channel through gamma correction to simulate changes in nighttime lighting conditions.

[0021] Further preferably, the method further includes randomly adding Poisson noise or Gaussian noise to the original training set X, wherein the Poisson noise satisfies a peak value of 10 and the Gaussian noise satisfies a standard deviation σ= Where B is a uniform distribution between (0,1).

[0022] Preferably, during the feature recognition process of the image sample data in the training set by the ReID network, multiple samples are taken from the initial simple distribution of the standard flow Θ.

[0023] In a second aspect, the present invention further proposes a system for target re-identification at night, the system comprising:

[0024] A training module, used to train the standardized flow module;

[0025] A preprocessing module is used to generate the low-light training set X by preprocessing through a standardized flow module l ;

[0026] The acquisition module is used to collect and acquire the training set, which includes the original training set X and the low-light training set X l ;

[0027] The feature recognition module is used to perform feature recognition on the image sample data in the training set through the ReID network.

[0028] In a third aspect, an embodiment of the present invention provides an electronic device comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0029] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0030] Compared with the prior art, the present invention has the following advantages:

[0031] (1) The technical solution of the present invention addresses the adverse interference caused by nighttime image acquisition. A standardized flow module is introduced outside the normal recognition network to perform low-light enhancement on the original acquired images. This module effectively improves the recognition capability of the re-identification network for low-light images, providing an effective technical solution for achieving target re-identification under nighttime conditions.

[0032] (2) The present invention proposes an end-to-end pedestrian re-identification network suitable for nighttime, introduces a standardized flow module with stronger modeling capabilities for low-light conditions, and organically embeds this module into the training process of the entire model; during the deployment process, the module can be flexibly selected according to the scene whether to use it. For example, under normal lighting conditions during the day, the features extracted by the re-ID network can be directly used; and under poor shooting conditions during the day (such as shadows) or at night, the features can be additionally enhanced to improve the recognition rate.

[0033] (3) The present invention obtains low-light data and the original data set, and adds additional low-light data to provide additional supervision information for the entire network. The addition of additional supervision information under a unified framework makes the entire network training more stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated into and constitute a part of this specification. The accompanying drawings illustrate the embodiments and, together with the description, serve to explain the principles of the present invention. Other embodiments and many of the expected advantages of the embodiments will be readily apparent as they become better understood by reference to the following detailed description. The elements of the drawings are not necessarily to scale with respect to each other. Like reference numerals designate corresponding similar parts.

[0035] Figure 1 is a diagram of an exemplary device architecture in which an embodiment of the present invention may be applied;

[0036] Figure 2 1 is a flow chart of a method for target re-identification at night according to an embodiment of the present invention;

[0037] Figure 3 Schematic diagram of the structure of a system for target re-identification at night according to an embodiment of the present invention;

[0038] Figure 4 It is a schematic structural diagram of a computer device suitable for implementing an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In the following detailed description, reference is made to the accompanying drawings, which form a part of the detailed description and are illustrated by illustrative specific embodiments in which the present invention may be practiced. To this end, directional terms, such as "top," "bottom," "left," "right," "up," "down," etc., are used with reference to the orientation of the figures being described. Because the components of the embodiments may be positioned in several different orientations, directional terms are used for illustrative purposes and are in no way limiting. It should be understood that other embodiments may be utilized or logical changes may be made without departing from the scope of the present invention. Therefore, the following detailed description should not be adopted in a limiting sense, and the scope of the present invention is defined by the appended claims.

[0040] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0041] Figure 1 An exemplary system architecture 100 is shown to which a method for processing information or an apparatus for processing information according to an embodiment of the present invention can be applied.

[0042] like Figure 1As shown, system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. Network 104 is a medium for providing communication links between terminal devices 101, 102, 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0043] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0044] The terminal devices 101 , 102 , and 103 may be various electronic devices with communication functions, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0045] The server 105 may be a server that provides various services, such as a background information processing server that processes verification request information sent by the terminal devices 101, 102, and 103. The background information processing server may analyze and perform other processing on the received verification request information and obtain a processing result (e.g., verification success information indicating that the verification request is a legitimate request).

[0046] It should be noted that the method for processing information provided in the embodiment of the present invention is generally executed by the server 105, and accordingly, the device for processing information is generally set in the server 105. In addition, the method for sending information provided in the embodiment of the present invention is generally executed by the terminal devices 101, 102, and 103, and accordingly, the device for sending information is generally set in the terminal devices 101, 102, and 103.

[0047] It should be noted that the server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (for example, to provide distributed services), or as a single software program or multiple software modules, without specific limitation herein.

[0048] Person re-identification (Person re-identification) uses image processing technology to determine whether a pedestrian captured by one camera appears in other cameras. This allows the pedestrian's movement path to be mapped and cross-camera tracking to be achieved. However, in low-light (nighttime) conditions, images captured by surveillance cameras often suffer from numerous issues such as color shift, low contrast, noise, and loss of detail. This image quality significantly reduces the recognition performance of existing object re-identification methods.

[0049] Normalizing flow transforms simple probability distributions into complex distributions through a series of reversible and differentiable mappings, providing a solution that can effectively and flexibly fit arbitrary distributions.

[0050] To address the problem of nighttime object re-identification, we need to address the significant interference that low-light environments can cause to image acquisition. The technical solution of this invention introduces a normalized flow module in addition to the normal recognition network, which effectively improves the re-identification network's ability to recognize low-light images.

[0051] Firstly, Figure 2 The embodiment of the present invention discloses a method for target re-identification at night, such as Figure 2 As shown, the method includes the following steps:

[0052] S101, training the standardized flow module;

[0053] In a specific embodiment, the normalized flow module is based on the low-light training set X l The low-light samples in the training data are used as output to effectively map low-light images to normal exposure images. During the first N epochs of training, the backbone network parameters are frozen, and only the normalization stream is trained first. When the current epoch exceeds N, the learning rate of the normalization stream module is adjusted to 1 / 10 of that of the backbone network. In this embodiment, N = 10.

[0054] Among them, through the sample pair (X l ,X) to measure the standard flow Θ, the loss function of the standard flow module is as follows:

[0055] L(X l ,X)=-logf flow (X l |X)

[0056] Among them, f flow (X l |X) represents the conditional probability density function of X.

[0057] Furthermore, the low-light training set Xl The original training set X is generated by adjusting the pixel values ​​of each channel through gamma correction to simulate the changes in nighttime lighting conditions. Poisson noise or Gaussian noise is randomly added to the original training set X, where Poisson noise meets the peak value of 10 and Gaussian noise meets the standard deviation Where B is a uniform distribution between (0,1).

[0058] S102, further preprocessing is performed through the standardized flow module to generate the low-light training set X l ;

[0059] S103: Collect and obtain a training set, wherein the training set includes an original training set X and a low-light training set X. l ;

[0060] S104: Perform feature recognition on the image sample data in the training set through a ReID network.

[0061] Specifically, during the feature recognition process of the image sample data in the training set by the ReID network, multiple samples are sampled from the initial simple distribution of the standard flow Θ.

[0062] In a specific embodiment, the method comprises the following steps:

[0063] S11, low-light training set X l The original training set X is generated by adjusting the pixel values ​​of each channel through gamma correction to simulate the changes in night lighting conditions. In order to make the obtained picture look more natural, Poisson noise or Gaussian noise is randomly added to it in the experiment, where Poisson noise meets the peak value of 10 and Gaussian noise meets the standard deviation Where B is a uniform distribution between (0,1);

[0064] S12, standard flow Θ loss function definition. Θ is passed through the sample pair (X l ,X), so its loss function can be defined as follows: Θ refers to the standard flow, and the following formula describes the loss function of the standard flow module

[0065] L(X l ,X)=-logf flow (X l |X)

[0066] Among them, f flow (X l |X) represents the conditional probability density function of X.

[0067] S13, training of the normalized flow module. The normalized flow module is trained with low-light samples X lThe output is an effective mapping from low-light to normal-exposure images. To effectively utilize the normalization flow module, the backbone network parameters are frozen during the first N epochs of training, prioritizing training of the normalization flow. In this experiment, N = 10. When the current epoch exceeds N, the learning rate of this module is adjusted to 1 / 10 of that of the backbone network.

[0068] S14. In the process of extracting image features in the reid network, multiple sampling can be performed from the initial simple distribution of the standard flow Θ (equivalent to simulating different lighting conditions) to obtain more robust features, thereby further improving the accuracy of the entire recognition network.

[0069] The training set consists of the low-light data obtained in step S11 and the original dataset. The addition of low-light data provides additional supervisory information for the entire network. Adding this additional supervisory information within a unified framework makes the entire network training more stable.

[0070] In addition, the technical solution of the present invention proposes an end-to-end pedestrian re-identification network suitable for nighttime pedestrians. It introduces a standardized flow module with stronger modeling capabilities in low-light conditions and organically embeds this module into the training process of the entire model. During deployment, this module can be flexibly selected according to the scenario. For example, under normal daytime lighting conditions, the features extracted by the ReID network can be directly used. However, under poor daytime shooting conditions (such as shadows) or at night, the features can be additionally enhanced with reference to step S14 to improve the recognition rate. Therefore, it can be said that this is a universal ReID model with stronger light adaptability.

[0071] The technical solution of the present invention addresses the adverse interference of night-time image acquisition and introduces a standardized flow module to perform low-light enhancement on the original acquired images, providing an effective technical solution for achieving target re-identification under night-time conditions.

[0072] In a second aspect, an embodiment of the present invention further discloses a system for target re-identification at night, such as Figure 3 As shown, the system includes: a training module 31, a preprocessing module 32, an acquisition module 33 and a feature recognition module 34.

[0073] In a specific embodiment, the training module 31 is used to train the normalized flow module; the preprocessing module 32 is used to generate the low-light training set X by preprocessing the normalized flow module. l ; Acquisition module 33, for collecting and acquiring a training set, the training set includes the original training set X and the low light training set X l ; Feature recognition module 34, used to perform feature recognition on the image sample data in the training set through the ReID network.

[0074] Reference below Figure 4 , which shows an electronic device (eg Figure 1 A structural diagram of a computer device 400 (a server or terminal device as shown). Figure 4 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0075] like Figure 4 As shown, the computer device 400 includes a central processing unit (CPU) 401 and a graphics processing unit (GPU) 402, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 403 or the program loaded from the storage part 409 to the random access memory (RAM) 406. Various programs and data required for the operation of the device 400 are also stored in the RAM 404. The CPU 401, GPU 402, ROM 403 and RAM 404 are connected to each other via a bus 405. An input / output (I / O) interface 406 is also connected to the bus 405.

[0076] The following components are connected to the I / O interface 406: an input section 407 including a keyboard, a mouse, and the like; an output section 408 including a display such as a liquid crystal display (LCD), a speaker, and the like; a storage section 409 including a hard disk and the like; and a communication section 410 including a network interface card such as a LAN card or a modem. The communication section 410 performs communication processing via a network such as the Internet. A drive 411 may also be connected to the I / O interface 406 as needed. A removable medium 412, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like, is installed in the drive 411 as needed, so that a computer program read therefrom can be installed into the storage section 409 as needed.

[0077] In particular, according to an embodiment disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment disclosed in the present invention includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication portion 410, and / or installed from a removable medium 412. When the computer program is executed by a central processing unit (CPU) 401 and a graphics processing unit (GPU) 402, the above-mentioned functions defined in the method of the present invention are performed.

[0078] It should be noted that the computer-readable medium described in the present invention may be a computer-readable signal medium or a computer-readable medium, or any combination thereof. Computer-readable media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor devices, apparatuses, or components, or any combination thereof. More specific examples of computer-readable media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution device, apparatus, or component. In the present invention, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution apparatus, device, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical cable, RF, or any suitable combination thereof.

[0079] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0080] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the devices, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based device that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0081] The modules involved in the embodiments of the present invention may be implemented in software or hardware, and may also be set in a processor.

[0082] As another aspect, the present invention further provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently and not be incorporated into the electronic device. The computer-readable medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device: trains the normalized stream module; further pre-processes the normalized stream module to generate the low-light training set X l ; Collect and obtain a training set, which includes the original training set X and the low-light training set X l ; Perform feature recognition on the image sample data in the training set through the ReID network.

[0083] The above description is merely an illustration of the preferred embodiments of the present invention and the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the present invention.

Claims

1. A method for target re-identification at night, characterized in that: The method comprises the following steps: Train the normalized flow module; Training the standardized flow module includes: the standardized flow module is trained with a low-light training set X l The low-light samples in are used as output to obtain an effective mapping from low-light to normal exposure images; The method further includes: in the training of the first N epochs, the parameters of the backbone network are frozen, and only the normalized flow is trained first; when the current epoch>N, the learning rate of the normalized flow module is adjusted to 1 / 10 of the backbone network; through the sample pair (X l ,X) is used to measure the standard flow Θ, and the loss function of the standardized flow module is as follows: L(X l ,X)=-logf flow (X l |X), where f flow (X l |X) represents the conditional probability density function of X; The low-light training set X is further generated by preprocessing through the standardized flow module l ; Collect and obtain a training set, which includes an original training set X and a low-light training set X l ; The image sample data in the training set is subjected to feature recognition through the ReID network.

2. The method for target re-identification at night according to claim 1, characterized in that: The low-light training set X l The original training set X is generated by adjusting the pixel values ​​of each channel through gamma correction to simulate changes in nighttime lighting conditions.

3. The method for target re-identification at night according to claim 2, characterized in that: The method further includes randomly adding Poisson noise or Gaussian noise to the original training set X, wherein the Poisson noise satisfies a peak value of 10 and the Gaussian noise satisfies a standard deviation of Where B is a uniform distribution between (0,1).

4. The method for target re-identification at night according to claim 1, characterized in that: In the process of performing feature recognition on the image sample data in the training set through the ReID network, multiple samples are sampled from the initial simple distribution of the standard flow Θ.

5. A system for target re-identification at night, characterized in that: The system includes: A training module, used to train the standardized flow module; Training the standardized flow module includes: the standardized flow module is trained with a low-light training set X l The low-light samples in are used as output to obtain an effective mapping from low-light to normal exposure images; The method further includes: in the training of the first N epochs, the parameters of the backbone network are frozen, and only the normalized flow is trained first; when the current epoch>N, the learning rate of the normalized flow module is adjusted to 1 / 10 of the backbone network; through the sample pair (X l ,X) is used to measure the standard flow Θ, and the loss function of the standardized flow module is as follows: L(X l ,X)=-logf flow (X l |X), where f flow (X l |X) represents the conditional probability density function of X; A preprocessing module is used to generate the low-light training set X by preprocessing through a standardized flow module l ; The acquisition module is used to collect and acquire the training set, which includes the original training set X and the low-light training set X l ; The feature recognition module is used to perform feature recognition on the image sample data in the training set through the ReID network.

6. An electronic device comprising: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images