CN116579909A - Dynamic image quality adjusting method, image quality prediction model training method and remote server - Google Patents
Dynamic image quality adjusting method, image quality prediction model training method and remote server Download PDFInfo
- Publication number
- CN116579909A CN116579909A CN202210110368.8A CN202210110368A CN116579909A CN 116579909 A CN116579909 A CN 116579909A CN 202210110368 A CN202210110368 A CN 202210110368A CN 116579909 A CN116579909 A CN 116579909A
- Authority
- CN
- China
- Prior art keywords
- image quality
- remote server
- prediction model
- residual
- remote
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000005540 biological transmission Effects 0.000 claims description 10
- 230000000903 blocking effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
A dynamic image quality adjusting method of a remote graphic processor, a training method of an image quality prediction model and a remote server applying the same. The dynamic image quality adjusting method of the remote graphic processor of the remote server comprises the following steps. Obtaining a network residual bandwidth of the remote server. A video memory residual capacity of the remote server is obtained. A residual power of a local electronic device is obtained. The network residual bandwidth, the video memory residual capacity and the residual power are input into an image quality prediction model to obtain an image quality set value. The remote graphic processor is controlled to set a picture resolution according to the picture quality setting value so as to process the picture.
Description
Technical Field
The disclosure relates to an image quality adjustment method, a training method of a prediction model and a server using the same, and more particularly to a dynamic image quality adjustment method of a remote graphics processor, a training method of a prediction model and a remote server using the same.
Background
Computers for electronic contests and creation have a strong computing power of a graphics processor, but most of computers are table computers or heavy notebook computers. The portability of this type of computer is quite poor, which results in the user being unable to engage in electronic competition or creation outside.
Furthermore, today's high-performance graphics processors consume considerable power. Even if a user takes the computer out of the door, the computer can only be used for 1 hour at most.
In order to enable a lightweight and thin notebook computer to perform electronic contests and creation, a Remote graphics processing technology (Remote GPU) has been developed. The operation requirement of the remote graphic processor can be transmitted back to the remote server by the notebook computer. Complex graphics processing may be performed by a remote server.
Disclosure of Invention
The present disclosure relates to a dynamic image quality adjustment method for a remote graphics processor, a training method for an image quality prediction model, and a remote server using the same, which utilizes the image quality prediction model to comprehensively consider the network residual bandwidth of the remote server and the video memory residual capacity of the remote server so as to predict the most suitable image quality setting value. The situations of network bandwidth blocking, operation delay and the like can be effectively avoided, and the user experience is optimized.
According to an aspect of the present disclosure, a method for dynamic image quality adjustment of a Remote graphics processor (Remote GPU) of a Remote server is provided. The dynamic image quality adjusting method of the remote graphic processor of the remote server comprises the following steps. Obtaining a network residual bandwidth of the remote server. A video memory remaining capacity of the remote server is obtained (VRAM remaining capacity). A residual power of a local electronic device is obtained. The network residual bandwidth, the video memory residual capacity and the residual power are input into an image quality prediction model to obtain an image quality set value. The remote graphic processor is controlled to set a picture resolution according to the picture quality setting value so as to process the picture.
According to another aspect of the present disclosure, a training method for an image quality prediction model is provided. The training method of the image quality prediction model comprises the following steps. The image quality analysis model is initially trained with a network bandwidth remaining at a remote server, a video memory remaining capacity (VRAM remaining capacity) of the remote server, and a remaining power of a local electronic device. And predicting by using the image quality prediction model to obtain a plurality of image quality setting values. Several scores corresponding to these image quality settings are collected. The image quality prediction model is retrained based on these scores.
According to yet another aspect of the present disclosure, a remote server is presented. The remote server comprises a remote graphic processor, a video memory, a network transmission unit, a picture quality prediction model and a picture quality setting unit. The video memory has a video memory remaining capacity (VRAM remaining capacity). The network transmission unit is connected with a local electronic device. The network transmission unit has a network residual bandwidth. The image quality prediction model is used for receiving the network residual bandwidth, the residual capacity of the video memory and a residual capacity of the local electronic device so as to output an image quality set value. The image quality setting unit is used for controlling the remote graphic processor to set a picture resolution according to the image quality setting value so as to process the picture.
For a better understanding of the above and other aspects of the disclosure, reference is made to the following detailed description of embodiments, taken in conjunction with the accompanying drawings, in which:
drawings
FIG. 1 is a schematic diagram illustrating the operation of a remote server according to an embodiment.
FIG. 2 shows an image quality prediction model according to an embodiment.
FIG. 3 is a block diagram of a remote server according to one embodiment.
FIG. 4 is a flow chart of a method for adjusting dynamic image quality of a remote graphics processor of a remote server according to an embodiment.
FIG. 5 is a flowchart of a training method of an image quality prediction model according to an embodiment.
FIG. 6 shows a 5-time predicted picture.
Description of the reference numerals
100 remote server
110 remote graphics processor
120 video memory
130 network transmission unit
140 image quality prediction model
150 image quality setting unit
200 local electronic device
900 network
BT: residual quantity of electric power
BW: network remaining bandwidth
CM arithmetic instruction
FM, FM1, FM2m FM3, FM4, FM5: picture
QS, QSi, QS1, QS2, QS3, QS4, QS5: image quality set point
RM video memory residual capacity
S110, S120, S130, S140, S150, S210, S220, S230, S240 steps
SC, SCi, SC 1-SC 5 score
Detailed Description
Referring to fig. 1, a schematic diagram of an operation of a remote server 100 according to an embodiment is shown. The local electronic device 200 may upload the calculation command CM to the remote server 100 through the network 900. The operation command CM of the application program may be transmitted to the virglrender renderer of the remote server 100 through the VirGL socket (socket). The Remote server 100 uses a Remote graphics processor (Remote GPU) 110 and the image FM calculated by the video memory 120 to transmit the image FM back to the local electronic device 200 through the network 900. After the local electronic device 200 obtains the picture FM, the picture FM can be displayed.
In this way, the local electronic device 200 can be kept light and thin, which can not only meet the requirement of convenient carrying, but also smoothly complete complex graphic operation. However, referring to the following table one, the bandwidth resources occupied by the different frame resolutions of the return frame FM are tens of times different. If the resolution of the returned frame FM is too high, the network bandwidth may be blocked, or the operation of the remote server 100 may be delayed. Therefore, the present technology proposes artificial intelligence techniques to optimize the resolution of the picture, optimizing the user's experience.
List one
Referring to fig. 2, an image quality prediction model 140 according to an embodiment is shown. The image quality prediction model 140 of the present embodiment is, for example, a neural network model. A network remaining bandwidth BW of the remote server 100, a video memory remaining capacity (VRAM remaining capacity) RM of the remote server 100, and a remaining power BT of the local electronic device 200 are input to the image quality prediction model 140 to obtain an image quality set value QS. The image quality setting value QS is 720P, 1080P, 4K, for example.
Referring to fig. 3, a block diagram of a remote server according to an embodiment is shown. The remote server 100 includes the remote graphic processor 110, the video memory 120, a network transmission unit 130, the image quality prediction model 140 and an image quality setting unit 150. The functions of the elements are summarized as follows. The remote graphic processor 110 is used for graphic processing, such as a processing chip or a circuit board. The video memory 120 is used for temporarily storing the processed graphics. The network transmission unit 130 is configured to transmit data, such as a Wifi transmission module or an LTE transmission module. The image quality prediction model 140 is used to predict the suitable image quality set point QS, for example, a circuit board, a chip, a program, or a recording medium storing the program. The image quality setting unit 150 is used for setting image quality, such as a circuit, a circuit board, a chip, a program, or a recording medium storing the program. The present embodiment predicts the best-fit image quality set point QS through the image quality prediction model 140, effectively avoids the situations of network bandwidth blocking, operation delay, etc., and optimizes the user experience.
The following describes a dynamic image quality adjustment method using the image quality prediction model 140. Next, a training method of the image quality prediction model 140 will be described.
Referring to fig. 4, a flowchart of a dynamic image quality adjustment method of the remote graphics processor 110 of the remote server 100 according to an embodiment is shown. In step S110, the network remaining bandwidth BW of the remote server 100 is obtained. The network remaining bandwidth BW is typically continuously hopping. The remote server 100 may periodically detect the network remaining bandwidth BW. In an embodiment, the network remaining bandwidth BW may be the uploaded bandwidth. The remaining bandwidth BW of the network is, for example, a relative percentage or absolute value.
Next, in step S120, the video memory remaining capacity RM of the remote server 100 is obtained. The remote server 100 may periodically detect the remaining video memory RM. The video memory remaining capacity RM is, for example, a relative percentage or absolute value.
Then, in step S130, the remaining power BT of the local electronic device 200 is obtained. The residual amount BT of the local electronic device 200 is transferred to the remote server 100 through the network 900. The remaining amount BT is, for example, a relative percentage or an absolute value. The steps S110, S120, S130 may be performed synchronously or in a switching order. The execution order of steps S110, S120, S130 is not limiting to the present invention.
Then, in step S140, the network remaining bandwidth BW, the video memory remaining capacity RM, and the remaining power BT are input to the image quality prediction model 140 to obtain the image quality set value QS. The image quality setting value QS is transmitted to the image quality setting unit 150.
Next, in step S150, the image quality setting unit 150 controls the remote graphic processor 110 to set a screen resolution according to the image quality setting value QS, so as to perform a screen process. The far-end graphics processor 110 processes the picture FM with a set picture resolution. The image quality setting value QS is, for example, an absolute value, such as 720P, 1080P, 4K. In another embodiment, the image quality setting QS is, for example, a relative adjustment value, for example, an up-step and a down-step.
In one embodiment, the resolution of the picture may be adjusted progressively to accommodate the viewing comfort of the user.
Referring to fig. 5, a flowchart of a training method of the image quality prediction model 140 according to an embodiment is shown. In step S210, the image quality prediction model 140 is initially trained with the network remaining bandwidth BW of the remote server 100, the video memory remaining capacity RM of the remote server 100, and the remaining power BT of a local electronic device 200 in an off-line state. The true value (ground truth) of the initial training image quality set point QS can be customized by a researcher or set to several values most commonly used in past history. The image quality prediction model 140 formed after the initial training does not have high accuracy.
Next, in step S220, in an on-line state (on-line), prediction is performed with the image quality prediction model 140 to obtain a number of image quality setting values QSi. This step is, for example, steps S110 to S150 described above with reference to fig. 4.
Then, in step S230, the local electronic device 200 collects a plurality of scores SCi corresponding to the image quality setting values QSi. The score SCi is, for example, the user's satisfaction with the image quality of the screen FM, the network speed, the delay time, and the like. These scores SCi are transmitted to the remote server 100 via the network.
Referring to fig. 6, 5 predicted pictures FM1 to FM5 are shown. For these pictures FM1 to FM5, the user can make a rating to obtain scores SC1 to SC5. Alternatively, the scoring may be automatically given by the system (e.g., periodically observing user adjustments: not manually adjusted resolution, indicating user satisfaction; manually adjusted resolution, indicating user dissatisfaction).
Next, in step S240, the image quality prediction model 140 is retrained according to the scores SCi to obtain the final image quality prediction model 140. The retrained image quality prediction model 140 has high accuracy, can accurately predict the most suitable image quality set value QS, effectively avoids the conditions of network bandwidth blocking, operation delay and the like, and optimizes the experience of a user.
According to the above embodiment, the present embodiment predicts the best-fit image quality set value QS by comprehensively considering the network remaining bandwidth BW of the remote server 100 and the video memory remaining capacity RM of the remote server 100 through the image quality prediction model 140. The situations of network bandwidth blocking, operation delay and the like can be effectively avoided, and the user experience is optimized.
In summary, although the present disclosure has been described above with reference to the embodiments, it is not intended to limit the disclosure. Those of ordinary skill in the art will appreciate that various modifications and adaptations can be made without departing from the spirit and scope of the present disclosure. Accordingly, the scope of the present disclosure is defined by the appended claims.
Claims (13)
1. A dynamic image quality adjusting method of a remote graphic processor of a remote server is characterized by comprising the following steps:
obtaining a network residual bandwidth of the remote server;
obtaining a video memory residual capacity of the remote server;
obtaining a residual electric quantity of a local electronic device;
inputting the network residual bandwidth, the video memory residual capacity and the residual electric quantity into an image quality prediction model to obtain an image quality set value; and
the remote graphic processor is controlled to set a picture resolution according to the picture quality setting value so as to process the picture.
2. The method of claim 1, wherein the network bandwidth is detected by the remote server.
3. The method of claim 1, wherein the video memory residual capacity is detected by the remote server.
4. The method of claim 1, wherein the residual power is transmitted from the local electronic device to the remote server.
5. The method of claim 1, wherein the image quality prediction model is disposed in the remote server.
6. The method of claim 1, wherein the resolution of the image is adjusted gradually.
7. The training method of the image quality prediction model is characterized by comprising the following steps of:
performing initial training on the image quality prediction model by using a network residual bandwidth of a remote server, a video memory residual capacity of the remote server and a residual electric quantity of a local electronic device;
predicting by the image quality prediction model to obtain a multi-stroke image quality set value;
collecting a plurality of scores corresponding to the image quality set values; and
retraining the image quality prediction model according to the scores.
8. The method of claim 7, wherein each score is a degree of satisfaction of the image quality.
9. The method of claim 1, wherein each of the scores includes a picture quality satisfaction and a delay time satisfaction.
10. The method according to claim 1, wherein the image quality prediction model is installed in the remote server.
11. A remote server, comprising:
a remote graphics processor;
a video memory having a video memory residual capacity;
a network transmission unit connected with a local electronic device, the network transmission unit having a network residual bandwidth;
a image quality prediction model for receiving the network residual bandwidth, the video memory residual capacity and a residual power of the local electronic device to output an image quality set value; and
an image quality setting unit for controlling the remote graphic processor to set a picture resolution according to the image quality setting value so as to process the picture.
12. The remote server of claim 11, wherein the remaining power is transmitted from the local electronic device to the remote server.
13. The remote server according to claim 11, wherein the image quality setting unit controls the remote graphic processor to progressively adjust the resolution of the image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210110368.8A CN116579909A (en) | 2022-01-29 | 2022-01-29 | Dynamic image quality adjusting method, image quality prediction model training method and remote server |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210110368.8A CN116579909A (en) | 2022-01-29 | 2022-01-29 | Dynamic image quality adjusting method, image quality prediction model training method and remote server |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116579909A true CN116579909A (en) | 2023-08-11 |
Family
ID=87538280
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210110368.8A Pending CN116579909A (en) | 2022-01-29 | 2022-01-29 | Dynamic image quality adjusting method, image quality prediction model training method and remote server |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116579909A (en) |
-
2022
- 2022-01-29 CN CN202210110368.8A patent/CN116579909A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11941883B2 (en) | Video classification method, model training method, device, and storage medium | |
US20180341872A1 (en) | Methods and systems for cnn network adaption and object online tracking | |
CN110189246B (en) | Image stylization generation method and device and electronic equipment | |
CN107392189B (en) | Method and device for determining driving behavior of unmanned vehicle | |
WO2016160221A1 (en) | Machine learning of real-time image capture parameters | |
CN110637460A (en) | Visual quality preserving quantitative parameter prediction using deep neural networks | |
CN107529098A (en) | Real-time video is made a summary | |
CN103973969A (en) | Electronic device and image composition method thereof | |
CN108229673B (en) | Convolutional neural network processing method and device and electronic equipment | |
CN111260037B (en) | Convolution operation method and device of image data, electronic equipment and storage medium | |
CN110930296A (en) | Image processing method, device, equipment and storage medium | |
CN114339054B (en) | Method and device for generating photographing mode and computer readable storage medium | |
US10733481B2 (en) | Cloud device, terminal device, and method for classifying images | |
CN113192470A (en) | Screen adjusting method and device, storage medium and electronic equipment | |
EP3725084A1 (en) | Deep learning on image frames to generate a summary | |
EP4240001A1 (en) | Photographing method and apparatus, and electronic device | |
CN112333481A (en) | Video pushing method and device, server and storage medium | |
CN113838134B (en) | Image key point detection method, device, terminal and storage medium | |
CN111199540A (en) | Image quality evaluation method, image quality evaluation device, electronic device, and storage medium | |
WO2021124328A1 (en) | System and method of determining a region of interest in media | |
CN116579909A (en) | Dynamic image quality adjusting method, image quality prediction model training method and remote server | |
CN115605922A (en) | Input image size switchable network for adaptive runtime efficient image classification | |
TWI837574B (en) | Dynamic image quality adjustment method of remote gpu, training method of image quality prediction model, and remote server using the same | |
CN113038055B (en) | Image processing method and device and electronic equipment | |
CN114419322B (en) | Image instance segmentation method and device, electronic equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |