专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明申请

WO2023088562A1 AN AUTOENCODER FOR DATA COMPRESSION 审中-公开
公开(公告)号：WO2023088562A1
公开(公告)日：2023-05-25
申请号：PCT/EP2021/082306
申请日：2021-11-19
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： LI, Yun
IPC分类号： G06N3/04 , G06N3/08 , H04N19/00 , G06N7/00
摘要： A computer-implemented method of training an autoencoder for data compression is provided. The method comprises inputting training data into a neural network to generate a plurality of first latent variables, using a probability estimator to obtain probability mass functions for the plurality of latent variables, and determining a rate measure based on the probability mass functions and a plurality of coded latent variables. The plurality of coded latent variables is based on one-hot encoding of plurality of quantized latent variables obtained by quantizing the plurality of latent variables. A gradient of the rate measure is determined based on the probability mass functions and a plurality of approximately coded latent variables obtained by applying a differentiable function to the plurality of latent variables to approximate one-hot encoding. One or more parameters of the autoencoder is updated based on the rate measure and the gradient of the rate measure.

2. 发明申请

WO2021262053A1 METHOD AND SYSTEM FOR IMAGE COMPRESSING AND CODING WITH DEEP LEARNING 审中-公开
公开(公告)号：WO2021262053A1
公开(公告)日：2021-12-30
申请号：PCT/SE2020/050663
申请日：2020-06-25
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： LI, Yun
IPC分类号： H04N19/167 , G06N3/04 , H04N19/13 , H04N19/91 , H04N19/124 , H04N19/182 , H04N19/1883
摘要： An image processing system (100) and methods therein for compressing and coding an image and for optimizing its parameters are disclosed. The embodiments herein provide an improved system and simplified method with deterministic uniform quantization with integer levels based on Softmax function for image compression and coding. The embodiments herein produce exact discrete probability mass function for latent variables to be coded from side information. The embodiments herein enable training of the image processing system to minimize the bit rate through backpropagation at the same time. Moreover, the embodiments herein create the possibility to encode region of interest (ROI) areas during coding.

3. 发明申请

WO2021063505A1 OBJECT DETECTION 审中-公开
公开(公告)号：WO2021063505A1
公开(公告)日：2021-04-08
申请号：PCT/EP2019/076765
申请日：2019-10-02
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： LI, Yun , HUNT, Alexander , CALTENCO, Hector , BASTANI, Saeed
IPC分类号： G06K9/00
摘要： A method of detecting objects in a video comprises, for each image in the video, determining (78) a degree of correlation between the image and at least one preceding image in a sequence of images; and obtaining (80) a first measure of energy consumption required for performing an object detection process on the image, obtaining a second measure of energy consumption required for performing an object tracking process for tracking at least one object from a previous image, and comparing (82b) the first and second measures of energy. The object detection process (84) is performed on the image if either said degree of correlation is below a threshold level or said first measure of energy is lower than the second measure of energy, and otherwise the object tracking process (86) is performed on the image.

4. 发明申请

WO2020114582A1 IMPROVED OPTICAL SEE-THROUGH VIEWING DEVICE AND METHOD FOR PROVIDING VIRTUAL CONTENT OVERLAPPING VISUAL OBJECTS 审中-公开
公开(公告)号：WO2020114582A1
公开(公告)日：2020-06-11
申请号：PCT/EP2018/083523
申请日：2018-12-04
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： HUNT, Alexander , BASTANI, Saeed , KRISTENSSON, Andreas , LI, Yun
IPC分类号： G06T5/00 , G06T7/80
摘要： An Optical See-Through viewing device (100) comprising a controller (101) configured to control a display arrangement (110) comprising an image capturing device (112), wherein the controller (101) is further configured to: a) display virtual content (DVC, 115) overlapping at least one real-life object (RLO) viewable by a user's eye (E) through the Optical See-Through viewing device (100); b) capture a composite view of displayed virtual content (DVC, 115) overlapping real life object (RLO); c) determine captured virtual content (CVC) based on composite view; d) determine a difference (D) between captured virtual content (CVC) and the displayed virtual content (DVC, 115); e) determine modified virtual content (MDVC) based on the difference (D); and f) display the modified virtual content (MDVC).

5. 发明申请

WO2020104038A1 A COMPLEXITY GRADING SYSTEM AND METHOD FOR EVALUATING AND SELECTING AN ARTIFICIAL NEURAL NETWORK 审中-公开
公开(公告)号：WO2020104038A1
公开(公告)日：2020-05-28
申请号：PCT/EP2018/082312
申请日：2018-11-23
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： BASTANI, Saeed , KRISTENSSON, Andreas , HUNT, Alexander , LI, Yun
IPC分类号： G06N3/04
摘要： A complexity grading system (100), a rule mining system (200) for generating a set of rules for the complexity grading system (100), and methods performed in a device using the complexity grading system (100) for evaluating and selecting a neural network to run an input inference task are disclosed. A rule comprises layers' activation patterns as a premise of the rule, a complexity grade as a conclusion of the rule, and a confidence level indicating how certain the complexity grading system is about this rule.

6. 发明申请

WO2023066507A1 AN AUTOENCODER FOR DATA COMPRESSION 审中-公开
公开(公告)号：WO2023066507A1
公开(公告)日：2023-04-27
申请号：PCT/EP2021/083240
申请日：2021-11-26
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： LI, Yun
IPC分类号： G06N3/04 , G06N3/08 , G06N7/00 , G06N20/20 , H04N19/00
摘要： A computer-implemented method of training an autoencoder for data compression is provided. The method comprises inputting data into a neural network to generate first latent variables and determining a distortion measure based on comparing the data to a reconstruction of the data. The reconstruction of the data is based on first quantized latent variables obtained by quantizing the first latent variables. The probability estimator is used to, for respective first latent variables, obtain one or more parameters of respective mixture models representing a probability distribution of the respective first latent variable. A first rate measure is determined based on the mixture models and approximately quantized latent variables, in which the approximately quantized latent variables are obtained by applying a differentiable function to the first latent variables to approximate quantization. One or more parameters of the autoencoder are updated based on the distortion measure and the first rate measure.

7. 发明申请

WO2022263892A1 EXTENDED REALITY OVERLAY OPTIMIZATION VIA DISTRIBUTED COMPUTING 审中-公开
公开(公告)号：WO2022263892A1
公开(公告)日：2022-12-22
申请号：PCT/IB2021/055344
申请日：2021-06-16
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： MCLACHLAN, Paul , CALTENCO, Héctor , DAHLGREN, Fredrik , LI, Yun
IPC分类号： G06T7/20 , G06K9/00 , G06T19/00
摘要： A method of a user device to track objects in an extended reality (XR) environment includes capturing an image of a physical environment associated with the XR environment, sending the image, a first timestamp for the image, and object information for at least one object in the image to a tracking service, receiving an object identifier, object tracking information, and a second timestamp for the at least one object from the tracking service, performing time and motion compensation on the object tracking information to correlate the object tracking information and object identifier with the at least one object in the image of the physical environment, and generating or updating an XR overlay for the at least one object using time and motion compensated object tracking information.

8. 发明申请

WO2022073596A1 A SYSTEM AND METHOD TO MINIMIZE REPETITIVE FLASHES IN STREAMING MEDIA CONTENT 审中-公开
公开(公告)号：WO2022073596A1
公开(公告)日：2022-04-14
申请号：PCT/EP2020/078059
申请日：2020-10-07
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： LI, Yun , KRISTENSSON, Andreas , DAHLGREN, Fredrik , HUNT, Alexander , FRIDH, Dan
IPC分类号： H04N21/44 , H04N21/4402 , H04N21/4545 , H04N21/45 , H04N21/845
摘要： An image processing system (100) and method therein for processing streaming media content to reduce disturbing spatial and/or temporal phenomena before being displayed on a device display. The image processing system (100) receives media stream comprising sequences of video frames into a buffer (110) and analyzes the video frames during streaming. The analyzing (130) is performed either on completely decoded or partially decoded video frames within a Group of Picture (GOP) to detect disturbing spatial and/or temporal phenomena. If a disturbing spatial and/or temporal phenomenon is detected, the image processing system (100) processes (140) the decoded video frames to reduce the disturbing spatial and/or temporal phenomenon based on user settings of the device and displays the processed video frames on the display (150).

9. 发明申请

WO2020244780A1 IMPROVED OPTICAL SEE-THROUGH VIEWING DEVICE AND METHOD FOR CALIBRATING PROVISION OF VIRTUAL CONTENT OVERLAPPING VISUAL OBJECTS 审中-公开
公开(公告)号：WO2020244780A1
公开(公告)日：2020-12-10
申请号：PCT/EP2019/064981
申请日：2019-06-07
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： HUNT, Alexander , BASTANI, Saeed , KRISTENSSON, Andreas , LI, Yun
IPC分类号： G02B27/00 , G02B27/01
摘要： An Optical See-Through viewing device (100) comprising a controller (101) configured to control a display arrangement (110) comprising an image capturing device (112) and a reflective device (114), wherein the controller (101) is further configured to calibrate the Optical See-Through viewing device (100) by: a) adapting (600) the reflective device (114) to enable the image capturing device to capture an image of a user's eye (E); b) capturing (610) an image of a user's eye (E); c) determining (620) a location of the user's eye (E); d) determining (630) a displacement for adapting the display arrangement (110); e) adapting (640) at least a part of the display arrangement (110, 112, 111); and f) returning (650) the reflective device (114).

10. 发明申请

WO2020164698A1 LOAD DISTRIBUTION FOR A DISTRIBUTED NEURAL NETWORK 审中-公开
公开(公告)号：WO2020164698A1
公开(公告)日：2020-08-20
申请号：PCT/EP2019/053536
申请日：2019-02-13
申请人： TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
发明人： BASTANI, Saeed , LI, Yun , BERKEMAN, Anders , HENNINGSSON, Maria , KALANTARI, Ashkan
IPC分类号： G06N3/063 , G06N3/04
摘要： A method for dynamic load distribution for a distributed neural network is disclosed. The method comprises estimating (103), in a device of the neural network, an energy usage for processing at least one non-processed layer in the device, and estimating (106), in the device of 5 the neural network, an energy usage for transmitting layer output of at least one processed layer to a cloud service of the neural network for processing. The method further comprises comparing (107), in the device of the neural network, the estimated energy usage for processing the at least one non-processed layer in the device with the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service. The 10 method furthermore comprises determining (108) to process the at least one non-processed layer in the device when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is equal or greater than the estimated energy usage for processing the at least one non-processed layer, and determining (110) to transmit the layer output of the at least one processed layer to the cloud service for processing 15 subsequent layers when the estimated energy usage for transmitting the layer output of the at least one processed layer to the cloud service is less than the estimated energy usage for processing the at least one non-processed layer in the device. Corresponding computer program product, apparatus, cloud service assembly, and system are also disclosed.20

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式