-
公开(公告)号:US20230056947A1
公开(公告)日:2023-02-23
申请号:US17652805
申请日:2022-02-28
摘要: According to one embodiment, a learning apparatus includes a processing circuit. The processing circuit acquires a first training condition and a first model trained in accordance with the first training condition, sets a second training condition used to reduce a model size of the first model, different from the first training condition, in accordance with the second training condition and based on the first model, trains a second model whose model size is smaller than that of the first model, and in accordance with a third training condition that is not the same as the second training condition and complies with the first training condition, trains a third model based on the second model.
-
公开(公告)号:US20230007275A1
公开(公告)日:2023-01-05
申请号:US17941083
申请日:2022-09-09
IPC分类号: H04N19/159 , H04N19/126 , H04N19/186 , H04N19/94 , H04N19/105 , H04N19/176 , H04N19/70 , H04N19/172 , H04N19/136 , H04N19/463
摘要: According to an embodiment, an encoding device includes an index setting unit and an encoding unit. The index setting unit generates a common index in which reference indices of one or more reference images included in a first index and a second index are sorted in a combination so as not to include a same reference image in accordance with a predetermined scanning order. The first index representing a combination of the one or more reference images referred to by a first reference image. The second index representing a combination of the one or more reference images referred to by a second reference image. The encoding unit encodes the common index.
-
公开(公告)号:US20220284238A1
公开(公告)日:2022-09-08
申请号:US17461082
申请日:2021-08-30
发明人: Shuhei NITTA , Akiyuki TANIZAWA
摘要: According to one embodiment, a learning apparatus includes a processor. The processor determines, based on a data resolution of subject data obtained at a subject device, a plurality of data resolutions that differ from one another within a range covering the data resolution of the subject data, the data resolutions each indicating a corresponding amount of information per unit. The processor trains a scalable network with training samples corresponding to each of the plurality of data resolutions, the scalable network being a neural network adapted to change a data resolution of input data.
-
公开(公告)号:US20220138471A1
公开(公告)日:2022-05-05
申请号:US17463107
申请日:2021-08-31
摘要: According to one embodiment, a load estimation apparatus includes processing circuitry. The processing circuitry acquires sensor data from a measurement target, calculates one or more posture loads relating to one or more body parts of the measurement target based on the sensor data, and displays display data including a posture load graph in which the one or more posture loads are illustrated in time series.
-
公开(公告)号:US20210241172A1
公开(公告)日:2021-08-05
申请号:US17002820
申请日:2020-08-26
发明人: Takahiro TANAKA , Kosuke HARUKI , Ryuji SAKAI , Akiyuki TANIZAWA , Atsushi YAGUCHI , Shuhei NITTA , Yukinobu SAKATA
摘要: A machine learning model compression system according to an embodiment includes one or more hardware processors configured to: select a layer of a trained machine learning model in order from an output side to an input side of the trained machine learning model; calculate, in units of an input channel, a first evaluation value evaluating a plurality of weights included in the selected layer; sort, in ascending order or descending order, the first evaluation values each calculated in units of the input channel; select a given number of the first evaluation values in ascending order of the first evaluation values; and delete the input channels used for calculation of the selected first evaluation values.
-
公开(公告)号:US20210065344A1
公开(公告)日:2021-03-04
申请号:US16802274
申请日:2020-02-26
IPC分类号: G06T7/00
摘要: A defect inspection device includes at least one memory storing instructions and at least one processor. The at least one processor is configured to execute the instructions to acquire a first image of an inspection target created in a first creation method, acquire a second image obtained by photographing the inspection target, extract index data similar to the acquired first image with reference to a database, the index data being a third image created in the first creation method or a feature quantity obtained by the third image, the database including the index data associated with correct data that is used as a comparison target of the index data and is an image determined not to be defective in previous inspection, acquire the correct data associated with the extracted index data in the database, generate a reference image on the basis of the acquired correct data, and estimate a pixel associated with a defective position on the inspection target photographed in the second image by comparing the reference image with the second image.
-
公开(公告)号:US20210012228A1
公开(公告)日:2021-01-14
申请号:US16803376
申请日:2020-02-27
摘要: An inference apparatus according to an embodiment of the present disclosure includes a memory and a hardware processor coupled to the memory. The hardware processor is configured to: acquire at least one control parameter of second machine learning model, the second machine learning model having a size smaller than a size of a first machine learning model input to the inference apparatus; change the first machine learning model to the second machine learning model based on the at least one control parameter; and perform inference in response to input data by using the second machine learning model.
-
8.
公开(公告)号:US20190191175A1
公开(公告)日:2019-06-20
申请号:US16282967
申请日:2019-02-22
发明人: Akiyuki TANIZAWA , Takeshi Chujoh
IPC分类号: H04N19/463 , H04N19/51 , H04N19/105 , H04N19/44 , H04N19/182 , H04N19/61 , H04N19/186 , H04N19/91
CPC分类号: H04N19/463 , H04N19/105 , H04N19/182 , H04N19/186 , H04N19/44 , H04N19/45 , H04N19/51 , H04N19/61 , H04N19/91
摘要: According to an embodiment, an encoding device includes: an index setting unit sets an index that represents information of a reference image and a weighting factor; an index reconfiguring unit predicts a reference value of the weighting factor, wherein the reference value indicates a factor to be set if a difference of pixel value between a reference image and a target image to be encoded is less than or equal to a specific value; and an entropy encoding unit encodes a difference value between the weighting factor and the reference value.
-
9.
公开(公告)号:US20170164006A1
公开(公告)日:2017-06-08
申请号:US15433570
申请日:2017-02-15
发明人: Akiyuki TANIZAWA , Takeshi CHUJOH
IPC分类号: H04N19/61 , H04N19/44 , H04N19/105 , H04N19/186 , H04N19/182 , H04N19/91
CPC分类号: H04N19/463 , H04N19/105 , H04N19/182 , H04N19/186 , H04N19/44 , H04N19/45 , H04N19/51 , H04N19/61 , H04N19/91
摘要: According to an embodiment, an encoding device includes: an index setting unit sets an index that represents information of a reference image and a weighting factor; an index reconfiguring unit predicts a reference value of the weighting factor, wherein the reference value indicates a factor to be set if a difference of pixel value between a reference image and a target image to be encoded is less than or equal to a specific value; and an entropy encoding unit encodes a difference value between the weighting factor and the reference value.
-
10.
公开(公告)号:US20170163998A1
公开(公告)日:2017-06-08
申请号:US15438357
申请日:2017-02-21
发明人: Akiyuki TANIZAWA , Takeshi CHUJOH
IPC分类号: H04N19/513 , H04N19/15 , H04N19/61 , H04N19/105 , H04N19/13
CPC分类号: H04N19/13 , H04N19/105 , H04N19/147 , H04N19/15 , H04N19/176 , H04N19/182 , H04N19/184 , H04N19/186 , H04N19/196 , H04N19/44 , H04N19/463 , H04N19/513 , H04N19/547 , H04N19/577 , H04N19/61 , H04N19/625 , H04N19/70
摘要: According to an embodiment, an encoding device includes a deriving unit and an encoding unit. The deriving unit is configured to derive a first reference value based on fixed point precision representing roughness of a weighting factor that is used for multiplying a reference image. The encoding unit is configured to encode a first difference value that is a difference value between the weighting factor and the first reference value and the fixed point precision. The weighting factor is included in a first range of predetermined bit precision having the first reference value at approximate center. The first difference value is in the predetermined range.
-
-
-
-
-
-
-
-
-