公开(公告)号：US11936899B2

公开(公告)日：2024-03-19

申请号：US17200631

申请日：2021-03-12

Applicant: Lemon Inc.

Inventor： Na Zhang ,  Kai Zhang ,  Hongbin Liu ,  Li Zhang ,  Yue Wang

IPC: H04N19/513 ,  H04N19/105 ,  H04N19/137 ,  H04N19/176 ,  H04N19/46

CPC classification number: H04N19/513 ,  H04N19/105 ,  H04N19/137 ,  H04N19/176 ,  H04N19/46

Abstract: Embodiments of the present disclosure provide methods, apparatuses and computer storage media for video processing. One example method comprises determining, during a conversion between a current video block of a video and a bitstream of the video, at least one set of motion candidates for the current video block, each motion candidate comprising predicted motion information determined from neighboring video blocks of the current video block; determining, based on a template of the current video block and from the at least one set of motion candidates, a target motion candidate for the current video block by using a template matching cost rule; and performing the conversion based on the target motion candidate.

公开(公告)号：US20240086685A1

公开(公告)日：2024-03-14

申请号：US18463389

申请日：2023-09-08

Applicant: Beijing Youzhuju Network Technology Co., Ltd. ,  Lemon Inc.

Inventor： Xiaosong ZHOU ,  Qingliang CAI ,  Shu CHEN ,  Zhe WANG ,  Haiqian HE

IPC: G06N3/0464

CPC classification number: G06N3/0464

Abstract: A method, apparatus, device and storage medium for recommending information. The method includes determining, based on a set of feature representations of a plurality of features associated with information recommendation, a first set of weights indicating importance of the plurality of features. The method also includes determining a second set of weights based on the set of feature representations and the first set of weights. The method further includes recommending the information to a user based on the set of feature representations, the first set of weights and the second set of weights. The importance of respective features associated with the information recommendation can be accurately determined through this method, which further improves the effectiveness of information recommendation and improves the user experience.

公开(公告)号：US20240086362A1

公开(公告)日：2024-03-14

申请号：US18475725

申请日：2023-09-27

Applicant: Lemon Inc. ,  Beijing Volcano Engine Technology Co., Ltd.

Inventor： Hao Wang ,  Jiaxin Ou ,  Sheng Qiu ,  Yi Wang ,  Zhengyu Yang ,  Yizheng Jiao ,  Jingwei Zhang ,  Jianyang Hu ,  Yang Liu ,  Ming Zhao ,  Hui Zhang ,  Kuankuan Guo ,  Huan Sun ,  Yinlin Zhang

IPC: G06F16/13

CPC classification number: G06F16/134

Abstract: A key-value store and a file system are integrated together to provide improved operations. The key-value store can include a log engine, a hash engine, a sorting engine, and a garbage collection manager. The features of the key-value store can be configured to reduce the number of I/O operations involving the file system, thereby improving read efficiency, reducing write latency, and reducing write amplification issues inherent in the combined key-value store and file system.

公开(公告)号：US20240086213A1

公开(公告)日：2024-03-14

申请号：US18484962

申请日：2023-10-11

Applicant: Lemon Inc. ,  Beijing Volcano Engine Technology Co., Ltd.

Inventor： Hui Zhang ,  Fei Liu ,  Ping Zhou ,  Chul Lee ,  Bo Li ,  Shan Xiao

IPC: G06F9/455 ,  G06F9/50

CPC classification number: G06F9/455 ,  G06F9/5027 ,  G06F2209/509

Abstract: Methods and systems for emulating a hardware accelerator is provided. When executed by a computer, the platform includes a plurality of computational resources provided by the computer; a hardware emulator operated on a first computational resource of the plurality of computational resources; and an accelerator being emulated in the platform and operated on a second computational resource of the plurality of computational resources, the accelerator being configured to execute an offloading operation.

公开(公告)号：US20240078792A1

公开(公告)日：2024-03-07

申请号：US17929449

申请日：2022-09-02

Applicant: Lemon Inc.

Inventor： Shuo Cheng ,  Wanchun Ma ,  Linjie Luo

IPC: G06V10/774 ,  G06V10/764 ,  G06V10/776 ,  G06V10/82 ,  G06V10/96 ,  G06V40/16

CPC classification number: G06V10/774 ,  G06V10/764 ,  G06V10/776 ,  G06V10/82 ,  G06V10/96 ,  G06V40/171 ,  G06V40/174

Abstract: Systems and methods for multi-task joint training of a neural network including an encoder module and a multi-headed attention mechanism are provided. In one aspect, the system includes a processor configured to receive input data including a first set of labels and a second set of labels. Using the encoder module, features are extracted from the input data. Using a multi-headed attention mechanism, training loss metrics are computed. A first training loss metric is computed using the extracted features and the first set of labels, and a second training loss metric is computed using the extracted features and the second set of labels. A first mask is applied to filter the first training loss metric, and a second mask is applied to filter the second training loss metric. A final training loss metric is computed based on the filtered first and second training loss metrics.

公开(公告)号：US20240070525A1

公开(公告)日：2024-02-29

申请号：US17897697

申请日：2022-08-29

Applicant: Lemon Inc.

Inventor： Jiankai Sun ,  Xinlei Xu ,  Xin Yang ,  Yuanshun Yao ,  Chong Wang

IPC: G06N20/00 ,  G06F21/62

CPC classification number: G06N20/00 ,  G06F21/6245

Abstract: The present disclosure describes techniques of performing machine unlearning in a recommendation model. An unlearning process of the recommendation model may be initiated in response to receiving a request for deleting a fraction of user data from any particular user. The recommendation model may be pre-trained to recommend content to users based at least in part on user data. Values of entries in a matrix corresponding to the fraction of user data may be configured as zero. The matrix may comprise entries denoting preferences of users with respect to content items. Confidence values associated with the fraction of user data may be configured as zero to block influence of the fraction of user data on performance of the recommendation model. The unlearning process may be implemented by performing a number of iterations until the recommendation model has converged.

公开(公告)号：US11900545B2

公开(公告)日：2024-02-13

申请号：US17570304

申请日：2022-01-06

Applicant: Lemon Inc.

Inventor： Nathanael Schager ,  Zhe Huang ,  Kewei Chen ,  Nite Luo ,  Clement Boissiere

IPC: G06T19/00 ,  G06T7/11 ,  G06T15/04 ,  G06T17/20

CPC classification number: G06T19/00 ,  G06T7/11 ,  G06T15/04 ,  G06T17/20 ,  G06T2207/10016

Abstract: The present disclosure describes techniques for creating effects based on facial features. A mesh comprising a set of vertices based on detecting key points of at least one facial part may be created. A mask may be created based on rendering the mesh. A filter may be created by applying a function to a texture based on the mask. A portion of at least one image may be extracted based at least in part on the mask. The portion may correspond to the at least one facial part. An effect may be created on the extracted portion using the filter.

公开(公告)号：US20240037080A1

公开(公告)日：2024-02-01

申请号：US17877644

申请日：2022-07-29

Applicant: Lemon Inc. ,  Beijing Youzhuju Network Technology Co., Ltd.

Inventor： Viacheslav Dubeyko ,  Jian Wang

IPC: G06F16/22 ,  G06F16/29 ,  G06F16/248

CPC classification number: G06F16/2228 ,  G06F16/29 ,  G06F16/248

Abstract: A first combined key may be generated based on a geographic location, a first time, and a first user that are associated with a first event. The first combined key and first data indicating the first event may be stored in a database, the first combined key configured to identify the first data. A second combined key may be generated based on the geographic location, a second time, and a second user that are associated with a second event. The second combined key and second data indicating the second event may be stored in the database, the second combined key configured to identify the second data. A set of events associated with the geographic location and comprising the first event and the second event may be presented by retrieving the first data and the second data using the first combined key and the second combined key, respectively.

公开(公告)号：US11888913B2

公开(公告)日：2024-01-30

申请号：US17721049

申请日：2022-04-14

Applicant: Lemon Inc.

Inventor： Ye-kui Wang

IPC: H04L65/65 ,  H04L67/02 ,  H04N21/84 ,  H04N21/435

CPC classification number: H04L65/65 ,  H04L67/02 ,  H04N21/435 ,  H04N21/84

Abstract: A mechanism for processing video data is disclosed. An essential property (EssentialProperty) of a Dynamic Adaptive Streaming over Hypertext transfer protocol (DASH) representation is determined. The EssentialProperty indicates that the representation is an external stream representation (ESR). A conversion is performed between a visual media data and a media presentation based on the based on the ESR.

公开(公告)号：US20240028566A1

公开(公告)日：2024-01-25

申请号：US18475746

申请日：2023-09-27

Applicant: Lemon Inc. ,  Beijing Volcano Engine Technology Co., Ltd.

Inventor： Sheng Qiu ,  Hao Wang ,  Zhengyu Yang ,  Yizheng Jiao ,  Jianyang Hu ,  Yang Liu ,  Jiaxin Ou ,  Huan Sun ,  Yinlin Zhang

IPC: G06F16/18 ,  G06F16/178 ,  G06F11/14

CPC classification number: G06F16/1815 ,  G06F16/178 ,  G06F16/1847 ,  G06F11/1407

Abstract: A file system particular for use with key-value stores is provided. The file system can operate in a user space instead of a kernel space. The file system can be an append-only file system. The file system can support use of solid state drives (SSDs) for storage, including zoned SSDs. The file system can include a file manager, a metadata manager, a task scheduler, a space allocator, and a collaborator for collaborating with a key-value store.