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. 发明公开

US20240046100A1 Apparatus, method and computer program for decoding neural network parameters and apparatus, method and computer program for encoding neural network parameters using an update model 审中-公开
公开(公告)号：US20240046100A1
公开(公告)日：2024-02-08
申请号：US18380551
申请日：2023-10-16
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Paul HAASE , Heiner KIRCHHOFFER , Daniel BECKING , Gerhard TECH , Karsten MUELLER , Wojciech SAMEK , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： G06N3/082
CPC分类号： G06N3/082
摘要： Embodiments according to the invention comprise an apparatus for decoding neural network parameters, which define a neural network. The apparatus may, optionally, be configured to obtain, e.g. to decode, parameters of a base model, e.g. NB, of the neural network which define one or more layers, e.g. base layers, of the neural network. Furthermore, the apparatus is configured to decode an update model, e.g. NU1 to NUK, which defines a modification of one or more layers, e.g. base layers, of the neural network, and the apparatus is configured modify parameters of a base model of the neural network using the update model, in order to obtain an updated model, e.g. designated as “new model” comprising new model layers LNkj. Moreover, the apparatus is configured to evaluate a skip information, e.g. a skip_row_flag and/or a skip_column_flag, indicating whether a sequence, e.g. a row, or a column or a block, of parameters of the update model is zero or not.

2. 发明申请

US20220222541A1 Neural Network Representation Formats 有权
公开(公告)号：US20220222541A1
公开(公告)日：2022-07-14
申请号：US17711569
申请日：2022-04-01
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Stefan MATLAGE , Paul HAASE , Heiner KIRCHHOFFER , Karsten MUELLER , Wojciech SAMEK , Simon WIEDEMANN , Detlev MARPE , Thomas SCHIERL , Yago SÁNCHEZ DE LA FUENTE , Robert SKUPIN , Thomas WIEGAND
IPC分类号： G06N3/10 , G06N3/08 , H03M7/30
摘要： Data stream having a representation of a neural network encoded thereinto, the data stream including serialization parameter indicating a coding order at which neural network parameters, which define neuron interconnections of the neural network, are encoded into the data stream.

3. 发明申请

US20210314618A1 INTRA PREDICTIONS USING LINEAR OR AFFINE TRANSFORMS WITH NEIGHBOURING SAMPLE REDUCTION 有权
公开(公告)号：US20210314618A1
公开(公告)日：2021-10-07
申请号：US17351017
申请日：2021-06-17
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Jonathan PFAFF , Philipp HELLE , Michael SCHAEFER , Roman RISCHKE , Tobias HINZ , Philipp MERKLE , Bjoern STALLENBERGER , Martin WINKEN , Mischa SIEKMANN , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： H04N19/61 , H04N19/176 , H04N19/184 , H04N19/44
摘要： Techniques for encoding/decoding of video signals are described, e.g. implemented in decoders, encoder, methods and non-transitory storage units storing instructions to perform the methods.
A decoder or encoder may be configured to predict a predetermined block of the picture using a plurality of neighbouring samples by reducing the plurality of neighbouring samples to obtain a reduced set of samples values lower, in number of samples, than compared to the plurality of neighbouring samples, subjecting the reduced set of sample values to a linear or affine linear transformation to obtain predicted values for predetermined samples of the predetermined block.

4. 发明申请

US20140362925A1 TRANSFORM COEFFICIENT CODING 有权
标题翻译：变换系数编码
公开(公告)号：US20140362925A1
公开(公告)日：2014-12-11
申请号：US14335439
申请日：2014-07-18
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Tung NGUYEN , Heiner KIRCHHOFFER , Detlev MARPE
IPC分类号： H04N19/60 , H04N19/91
CPC分类号： H04N19/60 , H03M7/4018 , H03M7/4075 , H04N19/91
摘要： An idea used herein is to use the same function for the dependency of the context and the dependency of the symbolization parameter on previously coded/decoded transform coefficients. Using the same function—with varying function parameter—may even be used with respect to different transform block sizes and/or frequency portions of the transform blocks in case of the transform coefficients being spatially arranged in transform blocks. A further variant of this idea is to use the same function for the dependency of a symbolization parameter on previously coded/decoded transform coefficients for different sizes of the current transform coefficient's transform block, different information component types of the current transform coefficient's transform block and/or different frequency portions the current transform coefficient is located within the transform block.
摘要翻译：这里使用的思想是使用相同的功能来使上下文的依赖性和符号化参数对先前编码/解码的变换系数的依赖性。使用具有变化函数参数的相同函数 - 甚至可以在变换系数被空间地布置在变换块中的情况下针对变换块的不同变换块大小和/或频率部分使用。该想法的另一变型是使用相同的功能来使符号化参数对先前编码/解码的变换系数的依赖性用于当前变换系数的变换块的不同大小，当前变换系数的变换块的不同信息分量类型和/ 或不同频率部分，当前变换系数位于变换块内。

5. 发明申请

US20130279577A1 PICTURE CODING SUPPORTING BLOCK MERGING AND SKIP MODE 有权
标题翻译：图像编码支持块合并和跳过模式
公开(公告)号：US20130279577A1
公开(公告)日：2013-10-24
申请号：US13875779
申请日：2013-05-02
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Heiko SCHWARZ , Heiner KIRCHHOFFER , Philipp HELLE , Simon OUDIN , Jan STEGEMANN , Benjamin BROSS , Detlev MARPE , Thomas WIEGAND
IPC分类号： H04N7/32
CPC分类号： H04N19/52 , H04N19/172 , H04N19/176 , H04N19/184 , H04N19/50 , H04N19/70
摘要： A coding efficiency increase is achieved by using a common signalization within the bitstream with regard to activation of merging and activation of the skip mode. One possible state of one or more syntax elements within the bitstream may signalize for a current sample set of a picture that the sample set is to be merged and has no prediction residual encoded and inserted into the bitstream. A common flag may signalize whether the coding parameters associated with a current sample set are to be set according to a merge candidate or to be retrieved from the bitstream, and whether the current sample set of the picture is to be reconstructed based on a prediction signal depending on the coding parameters associated with the current sample set, without any residual data, or to be reconstructed by refining the prediction signal depending on the coding parameters associated with the current sample set by means of residual data within the bitstream.
摘要翻译：通过在比特流中的共同信号化方面关于激活合并和激活跳过模式来实现编码效率提高。比特流内的一个或多个语法元素的一个可能状态可以针对样本集合被合并并且没有编码并插入到比特流中的预测残差的图像的当前采样集来信号通知。公共标志可以用信号通知根据合并候选来设置与当前采样组相关联的编码参数，还是要从比特流检索，以及是否根据预测信号来重新构建图像的当前采样集取决于与当前样本集相关联的编码参数，没有任何残差数据，或者根据与比特流内的剩余数据相关联的与当前采样集相关联的编码参数，通过精简预测信号来重构。

6. 发明公开

US20240155126A1 PARTITION-BASED INTRA-CODING CONCEPT 审中-公开
公开(公告)号：US20240155126A1
公开(公告)日：2024-05-09
申请号：US18525275
申请日：2023-11-30
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Santiago DE LUXÁN HERNÁNDEZ , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： H04N19/119 , H04N19/11 , H04N19/12 , H04N19/129 , H04N19/174 , H04N19/176 , H04N19/46 , H04N19/593 , H04N19/61
CPC分类号： H04N19/119 , H04N19/11 , H04N19/12 , H04N19/129 , H04N19/174 , H04N19/176 , H04N19/46 , H04N19/593 , H04N19/61
摘要： Block-based coding of a picture is rendered more effective by providing an intra-prediction coding concept according to which a certain block of the picture is intra-prediction coded using a certain intra-coding mode by partitioning the predetermined block into partitions along a certain dimension with the number of partitions being greater than two and/or the partitions being one sample wide along the certain dimension with the partitions being, for reconstruction purposes, sequentially subject to spatial prediction using the intra-prediction coding mode signaled for the certain block followed by correcting the thus obtained predictor using a prediction residual so that for preceding partitions a reconstruction of the samples is available to the decoder at the time of processing the next, then current, partition. Signaling overhead with respect to the partitioning may be left off or may be kept low.

7. 发明公开

US20240046093A1 DECODER, ENCODER, CONTROLLER, METHOD AND COMPUTER PROGRAM FOR UPDATING NEURAL NETWORK PARAMETERS USING NODE INFORMATION 审中-公开
公开(公告)号：US20240046093A1
公开(公告)日：2024-02-08
申请号：US18487885
申请日：2023-10-16
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Heiner KIRCHHOFFER , Karsten MÜLLER , Paul HAASE , Daniel BECKING , Gerhard TECH , Wojciech SAMEK , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： G06N3/08
CPC分类号： G06N3/08
摘要： Disclosed is a decoder for decoding parameters of a neural network, configured to obtain a plurality of neural network parameters of the neural network on the basis of an encoded bitstream, to obtain, e.g. to receive; e.g. to extract from an encoded bitstream, a node information describing a node of a parameter update tree, wherein the node information has a parent node identifier, which is, for example, a unique parent node identifier, for example an integer number, a string, and/or a cryptographic hash, and wherein the node information has a parameter update information, e.g. one or more update instructions, for example a difference signal between initial neural network parameters and a newer version thereof, e.g. corresponding to a child node of the update tree, and to derive one or more neural network parameters using parameter information of a parent node identified by the parent node identifier and using the parameter update information.

8. 发明公开

US20230238982A1 ARITHMETIC ENCODER FOR ARITHMETICALLY ENCODING AND ARITHMETIC DECODER FOR ARITHMETICALLY DECODING A SEQUENCE OF INFORMATION VALUES, METHODS FOR ARITHMETICALLY ENCODING AND DECODING A SEQUENCE OF INFORMATION VALUES AND COMPUTER PROGRAM FOR IMPLEMENTING THESE METHODS 审中-公开
公开(公告)号：US20230238982A1
公开(公告)日：2023-07-27
申请号：US18296055
申请日：2023-04-05
申请人： FRAUNHOFER-GESELLSCHAFT ZUR FOERDERUNG DER ANGEWANDTEN FORSCHUNG E.V.
发明人： Paul HAASE , Heiner KIRCHHOFFER , Karsten MUELLER , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： H03M7/30 , H03M7/40
CPC分类号： H03M7/6023 , H03M7/4006
摘要： The invention describes an encoding scheme for arithmetically encoding a sequence of information values into an arithmetic coded bitstream using providing the bitstream with entry point information allowing for resuming arithmetic decoding the bitstream from a predetermined entry point onward. A respective decoding scheme is also provided. These encoding and decoding schemes provide more efficient encoding concept in view of the decoding speed.

9. 发明申请

US20220132167A1 Block-Based Predictive Coding and Decoding of a Picture 有权
公开(公告)号：US20220132167A1
公开(公告)日：2022-04-28
申请号：US17465611
申请日：2021-09-02
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Jonathan PFAFF , Martin WINKEN , Christian RUDAT , Heiko SCHWARZ , Thomas WIEGAND , Detlev MARPE
IPC分类号： H04N19/65 , H04N19/13 , H04N19/176 , H04N19/60 , H04L65/60 , H04L65/80
摘要： A previously encoded or reconstructed version of a neighborhood of a predetermined block to be predicted is exploited so as to result into a more efficient predictive coding of the prediction block. In particular, a spectral decomposition of a region composed of this neighborhood and a first version of a predicted filling of the predetermined block results into a first spectrum which is subject to noise reduction and the thus resulting second spectrum may be subject to a spectral composition, thereby resulting in a modified version of this region including a second version of the predicted filling of the predetermined block. Owing to the exploitation of the already processed, i.e. encoded/reconstructed, neighborhood of the predetermined block, the second version of the predicted filling of the predetermined block tends to improve the coding efficiency.

10. 发明申请

US20210409708A1 SELECTIVE INTER-COMPONENT TRANSFORM (ICT) FOR IMAGE AND VIDEO CODING 有权
公开(公告)号：US20210409708A1
公开(公告)日：2021-12-30
申请号：US17472159
申请日：2021-09-10
申请人： Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V.
发明人： Christian HELMRICH , Christian RUDAT , Tung Hoang NGUYEN , Heiko SCHWARZ , Detlev MARPE , Thomas WIEGAND
IPC分类号： H04N19/12 , H04N19/61 , H04N19/46 , H04N19/176
摘要： An encoder for encoding a plurality of components of an image content region of an image to be encoded is configured for obtaining the plurality of components representing the image content region; selecting an intercomponent transform from a set of intercomponent transforms; encoding the plurality of components using the selected intercomponent transform to obtain encoded components; and providing the encoded components.

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