Image processing method, device, equipment and storage medium

Abstract

The disclosed embodiments provide an image processing method, apparatus, device and storage medium. Input the original image into a feature extraction network to obtain feature data; input the feature data into a set encoder to obtain encoded feature data; input the encoded feature data into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training. The image processing method provided by the disclosed embodiments implements encoding and decoding of feature data based on a jointly trained feature extraction network, a set encoder and a set decoder, which can improve the effect of encoding and decoding of feature data.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of image processing technology, and in particular, to an image processing method, apparatus, device and storage medium. Background Art

[0002] End-to-end compression codec based on deep learning has been greatly developed in recent years, which includes visual tasks and compression tasks. In related technologies, visual tasks are implemented based on uncompressed features during training. In actual use, features are compressed and reconstructed to achieve visual tasks. The difference between the original features and the reconstructed features will lead to a decrease in the performance of visual tasks. Summary of the invention

[0003] The embodiments of the present disclosure provide an image processing method, apparatus, device and storage medium, which implement encoding and decoding of feature data based on a jointly trained feature extraction network, a set encoder and a set decoder, and can improve the effect of encoding and decoding of feature data.

[0004] In a first aspect, an embodiment of the present disclosure provides an image processing method, including:

[0005] Input the original image into the feature extraction network to obtain feature data;

[0006] Inputting the characteristic data into a set encoder to obtain encoded characteristic data;

[0007] The encoded feature data is input into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training.

[0008] In a second aspect, the present disclosure also provides an image processing device, including:

[0009] A feature data acquisition module is used to input the original image into the feature extraction network to obtain feature data;

[0010] A coding feature data acquisition module, used for inputting the feature data into a set encoder to obtain coding feature data;

[0011] A decoding feature data acquisition module is used to input the encoded feature data into a set decoder to obtain decoded feature data so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training.

[0012] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0013] one or more processors;

[0014] a storage device for storing one or more programs,

[0015] When the one or more programs are executed by the one or more processors, the one or more processors implement the image processing method as described in the embodiment of the present disclosure.

[0016] In a fourth aspect, the embodiments of the present disclosure further provide a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the image processing method as described in the embodiments of the present disclosure.

[0017] The disclosed embodiments disclose an image processing method, apparatus, device and storage medium. The original image is input into a feature extraction network to obtain feature data; the feature data is input into a set encoder to obtain encoded feature data; the encoded feature data is input into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training. The image processing method provided by the disclosed embodiments realizes encoding and decoding of feature data based on the jointly trained feature extraction network, the set encoder and the set decoder, which can improve the effect of encoding and decoding of feature data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0019] Figure 1 A schematic diagram of a flow chart of an image processing method provided by an embodiment of the present disclosure;

[0020] Figure 2 is a structural diagram of a joint model provided by an embodiment of the present disclosure;

[0021] Figure 3 is a structural diagram of a joint model provided by an embodiment of the present disclosure;

[0022] Figure 4 is a structural diagram of a joint model provided by an embodiment of the present disclosure;

[0023] Figure 5 is a principle diagram of image processing provided by an embodiment of the present disclosure;

[0024] Figure 6 is a structural schematic diagram of an image processing device provided by an embodiment of the present disclosure;

[0025] Figure 7 It is a structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0027] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0028] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0029] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0030] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0031] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0032] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0033] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.

[0034] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0035] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0036] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0037] Figure 1 A flowchart of an image processing method provided in an embodiment of the present disclosure is provided. The embodiment of the present disclosure is applicable to the situation where characteristic data of an image is encoded and decoded. The method can be executed by an image processing device, which can be implemented in the form of software and / or hardware. Optionally, it can be implemented by an electronic device, which can be a mobile terminal, a PC or a server, etc.

[0038] like Figure 1 As shown, the method includes:

[0039] S110, inputting the original image into a feature extraction network to obtain feature data.

[0040] Among them, the original image can be any image to be processed. The feature extraction network can be a neural network for extracting features from an image, and can be composed of multiple convolutional layers. In this embodiment, the feature extraction network can be any existing open source neural network model for feature extraction, which is not limited here. For example: it can be a visual geometry group (Visual Geometry Group, VGG) network, a deep residual neural network (Deep residual network, ResNet) or a pedestrian re-identification (Person re-identification, ReID) network, etc. In this embodiment, the feature extraction network can be a trained neural network.

[0041] The feature data may be data representing features extracted from the original image, and may be represented in the form of an array or a matrix. Specifically, the feature data may be obtained after the original image is input into a trained feature extraction network.

[0042] S120, inputting the characteristic data into the set encoder to obtain the encoded characteristic data.

[0043] Among them, the encoder can be a trained entropy encoder, and entropy coding can be understood as converting an image into a compressed code stream (also called a bit stream) for transmission or storage. Entropy coding can be coding that does not lose any information, that is, lossless coding. The principle of entropy coding can be understood as: first determine the prior probability of each feature in the output feature data, and use fewer bits to represent the features with larger prior probabilities, and use more bits to represent the features with smaller prior probabilities, so that the total number of bits after encoding is minimized. In this embodiment, the entropy coding method can be Shannon coding, Huffman coding, and arithmetic coding, which are not limited here.

[0044] Specifically, the feature data extracted from the original image is input into an entropy encoder to perform entropy encoding on the feature data, thereby obtaining encoded feature data, and then the encoded feature data is stored or transmitted for subsequent decoding. In this embodiment, an entropy encoder is used to encode the feature data, and ultra-high compression is achieved while ensuring lossless image, thereby saving memory when storing the encoded feature data, and saving bandwidth resources when transmitting the encoded feature data.

[0045] S130, inputting the encoded feature data into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data.

[0046] The feature extraction network, the setting decoder and the setting decoder are obtained through joint training. The joint training can be understood as: the feature extraction network, the setting decoder and the setting decoder form a complete joint model, the joint model is trained with sample images, and after the training is completed, the joint model is disassembled for subsequent application stages.

[0047] The decoder may be a trained entropy decoder, and entropy decoding may be understood as a process of reconstructing the encoded feature data. Specifically, the encoded feature data is input into the entropy decoder to reconstruct the encoded feature data, obtain the decoded feature data, and perform subsequent image processing based on the decoded feature data. The subsequent image processing may be target detection, etc., which is not limited here.

[0048] Optionally, the method of jointly training the feature extraction network, the setting decoder and the setting decoder can be: connecting the feature extraction network, the setting decoder and the setting decoder in sequence to obtain a joint model; inputting the sample image into the joint model to obtain a first loss function, a second loss function and a third loss function; training the joint model based on at least one of the first loss function, the second loss function and the third loss function.

[0049] Among them, the process of inputting the sample image into the joint model can be: first inputting the sample image into the feature extraction network, outputting the sample feature data, and then inputting the sample feature data into the set encoder in sequence, outputting the sample decoding data and the predicted prior probability.

[0050] Among them, the first loss function can be a ternary loss function, the second loss function can be a center loss function, and the third loss function can be an entropy loss function.

[0051] Specifically, the process of inputting the sample image into the joint model to obtain the first loss function, the second loss function and the third loss function can be: inputting the sample image into the feature extraction network to obtain sample feature data; determining the first loss function and the second loss function based on the sample feature data; inputting the sample feature data into the set encoder and the set decoder in sequence, and outputting the predicted prior probability information; and determining the third loss function based on the predicted prior probability information.

[0052] Among them, the ternary loss function can be determined by the sample feature data corresponding to multiple groups of ternary sample images, wherein a group of ternary sample images can be understood as 3 sample images, two sample images are of the same category, and the third sample image is of another category. For example: the three sample images are two positive sample images and one negative sample image. Assume: the two positive sample images are represented as x1 and x2 respectively, the negative sample image is represented as x3, and f(x) represents the sample feature data. Then the ternary loss function can be expressed as: Among them, N represents the number of triplets, i represents the i-th triple sample image group, and α is a hyperparameter.

[0053] Among them, the center loss function can be expressed as: Among them, m represents the number of sample feature data, f(x i ) represents the sample feature data corresponding to the current sample image, C i Indicates the feature center corresponding to the current sample image. The feature center can be represented by the average value of the sample feature data that has been acquired.

[0054] Among them, the entropy loss function can be determined by a priori probability. In this embodiment, the principle of entropy coding can be to first learn the priori probability distribution of feature data, and then encode the feature data based on the probability distribution. Therefore, it is necessary to train the entropy encoder and entropy decoder based on the entropy loss function determined by the prior probability. The calculation formula of the entropy loss function can be expressed as: Among them, P(i) represents the predicted prior probability of the i-th feature in the sample feature data.

[0055] For example, Figure 2 is a schematic diagram of the structure of the joint model in this embodiment, such as Figure 2 As shown, in the order from input to output, the joint model includes a feature extraction network, an entropy encoder and an entropy decoder in sequence. The sample image is input into the feature extraction network to obtain sample feature data; the ternary loss function and the center loss function are determined based on the sample feature data; the sample feature data is input into the entropy encoder and the entropy decoder in sequence, the predicted prior probability is output, and the entropy loss function is determined based on the predicted prior probability.

[0056] Optionally, the process of training the joint model based on at least one of the first loss function, the second loss function and the third loss function can be: reversely adjusting the parameters of the feature extraction network based on the first loss function and the second loss function; and reversely adjusting the setting decoder, the setting decoder and the feature extraction network based on the third loss function.

[0057] Among them, the process of reverse parameter adjustment of the feature extraction network based on the first loss function and the second loss function can be: weighted accumulation of the first loss function and the second loss function, and the weighted accumulated loss function is passed from the output end of the feature extraction network to adjust the parameters of the feature extraction network. The process of reverse parameter adjustment of the setting decoder, the setting decoder and the feature extraction network based on the third loss function can be: the third loss function is passed from the output end of the setting decoder, and then passed to the feature extraction network after passing through the setting encoder to adjust the setting decoder, the setting decoder and the feature extraction network. In this embodiment, training the joint model based on multiple loss functions can improve the accuracy of the model.

[0058] Optionally, the joint model further includes an object detection network; the object detection network is connected to the set decoder. The training method of the joint model can be: inputting the sample feature data into the set encoder and the set decoder in sequence, and outputting the sample decoded feature data; inputting the sample decoded feature data into the object detection network to obtain the object detection result; determining the fourth loss function based on the object detection result; and training the joint model based on at least one of the first loss function, the second loss function, the third loss function, and the fourth loss function.

[0059] The fourth loss function may be a cross entropy loss function or a label smoothing loss function, etc., which is not limited here. The fourth loss function may be determined based on the object detection result by obtaining the true result of the sample image and determining the fourth loss function according to the object detection result and the true result.

[0060] In this embodiment, the process of training the joint model based on at least one of the first loss function, the second loss function, the third loss function and the fourth loss function can be: reverse parameter adjustment of the feature extraction network based on the first loss function and the second loss function; reverse parameter adjustment of the setting decoder, the setting decoder and the feature extraction network based on the third loss function; reverse parameter adjustment of the object detection network, the setting decoder, the setting decoder and the feature extraction network based on the fourth loss function.

[0061] Among them, the process of reverse parameter adjustment of the feature extraction network based on the first loss function and the second loss function can be: weighted accumulation of the first loss function and the second loss function, and the weighted accumulated loss function is passed from the output end of the feature extraction network to adjust the parameters of the feature extraction network. The process of reverse parameter adjustment of the setting decoder, the setting decoder and the feature extraction network based on the third loss function can be: the third loss function is passed from the output end of the setting decoder, and then passed to the feature extraction network after passing through the setting encoder, so as to adjust the setting decoder, the setting decoder and the feature extraction network. The process of reverse parameter adjustment of the object detection network, the setting decoder, the setting decoder and the feature extraction network based on the fourth loss function can be: the fourth loss function is passed from the output end of the object detection network, and then passed through the setting decoder and the setting encoder in turn to the feature extraction network, so as to adjust the object detection network, the setting decoder, the setting decoder and the feature extraction network. In this embodiment, the object detection network is added to the joint model for joint training with the feature extraction network, the setting decoder and the setting decoder, which can improve the training accuracy of the feature extraction network, the setting decoder and the setting decoder. Exemplary, Figure 3 is a schematic diagram of the structure of the joint model in this embodiment, such as Figure 3 As shown, in the order from input to output, the joint model includes a feature extraction network, an entropy encoder, an entropy decoder and an object detection network.

[0062] Optionally, the joint model also includes a quantization module; the quantization module is arranged between the feature extraction network and the set encoder; the sample feature data is input into the quantization module to obtain quantized sample feature data; accordingly, the process of determining the first loss function and the second loss function based on the sample feature data can be: determining the first loss function and the second loss function based on the quantized sample feature data.

[0063] The function of the quantization module is to quantize the sample feature data output by the feature extraction network. Quantization can be understood as the process of reducing the amount of feature data, for example, converting a floating point number into an integer, or reducing the number of digits after the decimal point of a floating point number. For example, Figure 4 is a schematic diagram of a joint model in this embodiment, such as Figure 4 As shown, a quantization module is also included between the feature extraction network and the upper encoder for quantizing the sample feature data.

[0064] Optionally, after training the joint model, the method further includes: splitting the joint model, and deploying the split feature extraction network, quantization module and setting encoder on a terminal device; and deploying the split setting decoder on another terminal device.

[0065] Specifically, after the joint model is split, the original image is input into the feature extraction network, quantization module and set encoder deployed on one of the terminal devices in sequence to obtain encoded feature data. The encoded feature data is transmitted to another terminal device deployed with a set decoder, so that the other terminal device decodes the encoded feature data by calling the set decoder to obtain decoded feature data.

[0066] Optionally, after inputting the original image into the feature extraction network to obtain feature data, it also includes: inputting the feature data into a quantization module to obtain quantized feature data; correspondingly, the feature data is input into a set encoder, and the method of obtaining encoded feature data can be: inputting the quantized feature data into the set encoder to obtain encoded feature data.

[0067] For example, Figure 5 is a schematic diagram of the image processing in this embodiment, such as Figure 5 As shown, the original image is sequentially input into the feature extraction network, quantization module and set encoder deployed on terminal device A to obtain encoded feature data. The encoded feature data is transmitted to terminal device B deployed with a set decoder, so that terminal device B decodes the encoded feature data by calling the set decoder to obtain decoded feature data.

[0068] The technical solution of the embodiment of the present disclosure is to input the original image into the feature extraction network to obtain feature data; input the feature data into the set encoder to obtain encoded feature data; input the encoded feature data into the set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained by joint training. The image processing method provided by the embodiment of the present disclosure implements encoding and decoding of feature data based on the jointly trained feature extraction network, the set encoder and the set decoder, which can improve the effect of encoding and decoding of feature data.

[0069] Figure 6 A schematic diagram of the structure of an image processing device provided by an embodiment of the present disclosure is shown in FIG. Figure 6 As shown, the device comprises:

[0070] Feature data acquisition module 610, used to input the original image into the feature extraction network to obtain feature data;

[0071] The coding feature data acquisition module 620 is used to input the feature data into the set encoder to obtain the coding feature data;

[0072] The decoding feature data acquisition module 630 is used to input the encoded feature data into the setting decoder to obtain the decoding feature data so as to perform image processing according to the decoding feature data; wherein the feature extraction network, the setting decoder and the setting decoder are obtained through joint training.

[0073] Optionally, the encoder is set to an entropy encoder and the decoder is set to an entropy decoder.

[0074] Optionally, it also includes: a joint model training module for:

[0075] Connect the feature extraction network, the setting decoder and the setting decoder in sequence to obtain a joint model;

[0076] Input the sample image into the joint model to obtain a first loss function, a second loss function and a third loss function;

[0077] The joint model is trained based on at least one of the first loss function, the second loss function, and the third loss function.

[0078] Optionally, the joint model training module is also used to:

[0079] Input the sample image into the feature extraction network to obtain sample feature data;

[0080] Determine a first loss function and a second loss function based on the sample feature data; wherein the first loss function is a ternary loss function, and the second loss function is a center loss function;

[0081] Input the sample feature data into the set encoder and the set decoder in sequence, and output the predicted prior probability;

[0082] A third loss function is determined based on the predicted prior probability; wherein the third loss function is an entropy loss function.

[0083] Optionally, the joint model training module is also used to:

[0084] Reversely adjust the parameters of the feature extraction network based on the first loss function and the second loss function;

[0085] Based on the third loss function, reverse parameter adjustment is performed on the setting decoder, setting decoder and feature extraction network.

[0086] Optionally, the joint model further includes an object detection network; the object detection network is connected to the set decoder; and the joint model training module is further used to:

[0087] Input the sample feature data into the set encoder and the set decoder in sequence, and output the sample decoded feature data;

[0088] Input the sample decoded feature data into the object detection network to obtain the object detection result;

[0089] determining a fourth loss function based on the object detection result;

[0090] The joint model is trained based on at least one of the first loss function, the second loss function, the third loss function, and the fourth loss function.

[0091] Optionally, the joint model further includes a quantization module; the quantization module is arranged between the feature extraction network and the setting encoder; the joint model training module is further used for:

[0092] Input the sample feature data into the quantization module to obtain quantized sample feature data;

[0093] A first loss function and a second loss function are determined based on the quantized sample feature data.

[0094] Optionally, after training the joint model, a joint model splitting module is further included, for:

[0095] Split the joint model and deploy the split feature extraction network, quantization module and setting encoder on a terminal device;

[0096] The split setting decoder is deployed on another terminal device.

[0097] Optionally, also include:

[0098] Input the feature data into the input quantization module to obtain quantized feature data;

[0099] Accordingly, the characteristic data is input into the set encoder to obtain the encoded characteristic data, including:

[0100] The quantized feature data is input into the set encoder to obtain the encoded feature data.

[0101] The image processing device provided in the embodiments of the present disclosure can execute the image processing method provided in any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0102] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present disclosure.

[0103] Figure 7 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure. Figure 7 , which shows an electronic device (eg, Figure 7 The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0104] like Figure 7 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An edit / output (I / O) interface 505 is also connected to the bus 504.

[0105] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0106] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains 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 through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0107] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0108] The electronic device provided by the embodiment of the present disclosure and the image processing method provided by the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0109] The embodiments of the present disclosure provide a computer storage medium on which a computer program is stored. When the program is executed by a processor, the image processing method provided by the above embodiments is implemented.

[0110] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0111] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0112] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0113] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device:

[0114] The above-mentioned computer-readable medium carries one or more programs. When the above-mentioned one or more programs are executed by the electronic device, the electronic device: inputs the original image into the feature extraction network to obtain feature data; inputs the feature data into the set encoder to obtain encoded feature data; inputs the encoded feature data into the set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training.

[0115] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, 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 separate software package, partially on the user's computer and partially on a remote computer, or entirely on a 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).

[0116] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square 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 square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0117] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a unit does not limit the unit itself in some cases. For example, the first acquisition unit may also be described as a "unit for acquiring at least two Internet Protocol addresses".

[0118] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0119] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0120] According to one or more embodiments of the present disclosure, there is provided an image processing method, comprising:

[0121] Input the original image into the feature extraction network to obtain feature data;

[0122] Inputting the characteristic data into a set encoder to obtain encoded characteristic data;

[0123] The encoded feature data is input into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set decoder and the set decoder are obtained through joint training.

[0124] Furthermore, the encoder is set to be an entropy encoder, and the decoder is set to be an entropy decoder.

[0125] Furthermore, the method of jointly training the feature extraction network, the setting decoder and the setting decoder is:

[0126] Connecting the feature extraction network, the setting decoder and the setting decoder in sequence to obtain a joint model;

[0127] Inputting the sample image into the joint model to obtain a first loss function, a second loss function and a third loss function;

[0128] The joint model is trained based on at least one of the first loss function, the second loss function and the third loss function.

[0129] Furthermore, the sample image is input into the joint model to obtain a first loss function, a second loss function and a third loss function, including:

[0130] Inputting the sample image into the feature extraction network to obtain sample feature data;

[0131] Determine a first loss function and a second loss function based on the sample feature data; wherein the first loss function is a ternary loss function, and the second loss function is a center loss function;

[0132] Inputting the sample feature data into the set encoder and the set decoder in sequence, and outputting a predicted prior probability;

[0133] A third loss function is determined based on the predicted prior probability; wherein the third loss function is an entropy loss function.

[0134] Further, training the joint model based on at least one of the first loss function, the second loss function, and the third loss function includes:

[0135] Performing reverse parameter adjustment on the feature extraction network based on the first loss function and the second loss function;

[0136] The setting decoder, the setting decoder and the feature extraction network are reversely adjusted based on the third loss function.

[0137] Furthermore, the joint model also includes an object detection network; the object detection network is connected to the setting decoder; the training method of the joint model is:

[0138] Inputting the sample characteristic data into the set encoder and the set decoder in sequence, and outputting sample decoded characteristic data;

[0139] Inputting the sample decoded feature data into the object detection network to obtain an object detection result;

[0140] determining a fourth loss function based on the object detection result;

[0141] Accordingly, training the joint model based on at least one of the first loss function, the second loss function, and the third loss function includes:

[0142] The joint model is trained based on at least one of the first loss function, the second loss function, the third loss function and the fourth loss function.

[0143] Furthermore, the joint model further comprises a quantization module; the quantization module is arranged between the feature extraction network and the setting encoder;

[0144] Inputting the sample characteristic data into the quantization module to obtain quantized sample characteristic data;

[0145] Accordingly, determining the first loss function and the second loss function based on the sample feature data includes:

[0146] A first loss function and a second loss function are determined based on the quantized sample feature data.

[0147] Furthermore, after training the joint model, the method further includes:

[0148] Splitting the joint model, and deploying the split feature extraction network, quantization module, and setting encoder on a terminal device;

[0149] The split setting decoder is deployed on another terminal device.

[0150] Furthermore, after the original image is input into the feature extraction network to obtain feature data, the method further includes:

[0151] Inputting the feature data into the quantization module to obtain quantized feature data;

[0152] Accordingly, the characteristic data is input into a setting encoder to obtain encoding characteristic data, including:

[0153] The quantized feature data is input into the set encoder to obtain the encoded feature data.

[0154] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0155] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0156] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. An image processing method, characterized in that: include: Input the original image into the feature extraction network to obtain feature data; Inputting the characteristic data into a set encoder to obtain encoded characteristic data; Inputting the encoded feature data into a set decoder to obtain decoded feature data, so as to perform image processing according to the decoded feature data; wherein the feature extraction network, the set encoder and the set decoder are obtained by joint training; The feature extraction network, the setting encoder and the setting decoder are obtained by training a joint model; wherein the joint model is obtained by connecting the feature extraction network, the setting encoder and the setting decoder; Among them, the joint model also includes an object detection network, and the joint model is obtained by connecting the feature extraction network, the setting encoder, the setting decoder and the object detection network in sequence; the training method of the joint model is: training the joint model based on at least one of the first loss function, the second loss function, the third loss function and the fourth loss function.

2. The method according to claim 1, characterized in that The encoder is set to be an entropy encoder, and the decoder is set to be an entropy decoder.

3. The method according to claim 2, characterized in that The method of jointly training the feature extraction network, the setting encoder and the setting decoder is: Connecting the feature extraction network, the set encoder and the set decoder in sequence to obtain a joint model; Inputting the sample image into the joint model to obtain a first loss function, a second loss function and a third loss function; The joint model is trained based on at least one of the first loss function, the second loss function and the third loss function.

4. The method according to claim 3, characterized in that Inputting the sample image into the joint model to obtain a first loss function, a second loss function and a third loss function includes: Inputting the sample image into the feature extraction network to obtain sample feature data; Determine a first loss function and a second loss function based on the sample feature data; wherein the first loss function is a ternary loss function, and the second loss function is a center loss function; Inputting the sample feature data into the set encoder and the set decoder in sequence, and outputting a predicted prior probability; A third loss function is determined based on the predicted prior probability; wherein the third loss function is an entropy loss function.

5. The method according to claim 3 or 4, characterized in that: Training the joint model based on at least one of the first loss function, the second loss function, and the third loss function includes: Performing reverse parameter adjustment on the feature extraction network based on the first loss function and the second loss function; The set encoder, the set decoder and the feature extraction network are reversely adjusted based on the third loss function.

6. The method according to claim 4, characterized in that The object detection network is connected to the setting decoder; the training method of the joint model is: Inputting the sample characteristic data into the set encoder and the set decoder in sequence, and outputting sample decoded characteristic data; Inputting the sample decoded feature data into the object detection network to obtain an object detection result; A fourth loss function is determined based on the object detection result.

7. The method according to claim 4 or 6, characterized in that: The joint model also includes a quantization module; the quantization module is arranged between the feature extraction network and the setting encoder; Inputting the sample characteristic data into the quantization module to obtain quantized sample characteristic data; Accordingly, determining the first loss function and the second loss function based on the sample feature data includes: A first loss function and a second loss function are determined based on the quantized sample feature data.

8. The method according to claim 7, characterized in that After training the joint model, the method further includes: Splitting the joint model, and deploying the split feature extraction network, quantization module, and setting encoder on a terminal device; The split setting decoder is deployed on another terminal device.

9. The method according to claim 8, characterized in that After the original image is input into the feature extraction network and feature data is obtained, it also includes: Inputting the feature data into the quantization module to obtain quantized feature data; Accordingly, the characteristic data is input into a setting encoder to obtain encoding characteristic data, including: The quantized feature data is input into the set encoder to obtain the encoded feature data.

10. An image processing device, characterized in that: include: A feature data acquisition module is used to input the original image into the feature extraction network to obtain feature data; A coding feature data acquisition module, used for inputting the feature data into a set encoder to obtain coding feature data; A decoding feature data acquisition module, used for inputting the encoding feature data into a set decoder to obtain decoding feature data, so as to perform image processing according to the decoding feature data; wherein the feature extraction network, the set encoder and the set decoder are obtained by joint training; The feature extraction network, the setting encoder and the setting decoder are obtained by training a joint model; wherein the joint model is obtained by connecting the feature extraction network, the setting encoder and the setting decoder; Among them, the joint model also includes an object detection network, and the joint model is obtained by connecting the feature extraction network, the setting encoder, the setting decoder and the object detection network in sequence; the training method of the joint model is: training the joint model based on at least one of the first loss function, the second loss function, the third loss function and the fourth loss function.

11. An electronic device, characterized in that: The electronic device comprises: 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 image processing method as described in any one of claims 1 to 9.

12. A storage medium comprising computer executable instructions, wherein the computer executable instructions are used to perform the image processing method according to any one of claims 1 to 9 when executed by a computer processor.

Images