公开(公告)号：US20240153255A1

公开(公告)日：2024-05-09

申请号：US18387515

申请日：2023-11-07

Applicant: NEC Corporation

Inventor： Tomokazu KANEKO ,  Soma SHIRAISHI ,  Ryosuke SAKAI

IPC: G06V10/776 ,  G06V10/764 ,  G06V10/77 ,  G06V20/40

CPC classification number: G06V10/776 ,  G06V10/764 ,  G06V10/7715 ,  G06V20/46

Abstract: A learning device includes a learning model for a still image. The learning model includes a mask generation means configured to generate a first object mask identifying an area in which an object exists in a still image, for each individual object. A first parameter including at least one parameter used for processing of generating the first object mask is adjusted based on a first loss. The first loss indicates a difference of the first object mask with respect to a second object mask identifying an area in which an object exists in a moving image including the still image, for each individual object.

公开(公告)号：US20240062048A1

公开(公告)日：2024-02-22

申请号：US18269790

申请日：2020-12-28

Applicant: NEC Corporation

Inventor： Ryosuke SAKAI

IPC: G06N3/047 ,  G06N3/084

CPC classification number: G06N3/047 ,  G06N3/084

Abstract: A learning device 1X includes a probabilistic inference result generation means 16X, a formatting means 17X, and a training means 18X. The probabilistic inference result generation means 16X is configured to generate a probabilistic inference result that is probabilistically generated for an input data. The formatting means 17X is configured to generate a formatted inference result obtained by formatting the probabilistic inference result. The training means 18X is configured to train a correction learning model that is a learning model configured to correct the formatted inference result, based on the input data, correct answer data corresponding to the input data, and the formatted inference result.

公开(公告)号：US20230162388A1

公开(公告)日：2023-05-25

申请号：US17916870

申请日：2020-04-13

Applicant: NEC Corporation

Inventor： Ryosuke SAKAI

IPC: G06T7/73 ,  G06T7/68

CPC classification number: G06T7/73 ,  G06T7/68 ,  G06T2207/20081

Abstract: The learning device 1A includes an acquiring means 23A, a conversion means 24A, and a learning means 25A. The acquiring means 23A is configured to acquire a combination of a first label that is a unique label for each feature point of an object and a feature point image in which a feature point corresponding to the first label is imaged. The conversion means 24A is configured to convert the first label to a second label that is set to a same label for feature points of the object with at least one of a congruence relation in appearance or a mirror symmetry relation to one another. The learning means 25A is configured to learn an inference engine based on the second label, the feature point image, and correct answer data regarding a position of the feature point, the inference engine being configured to perform an inference on the position of the feature point included in an image that is inputted to the inference engine.

公开(公告)号：US20220254136A1

公开(公告)日：2022-08-11

申请号：US17587065

申请日：2022-01-28

Applicant: NEC Corporation

Inventor： Tomokazu KANEKO ,  Katsuhiko TAKAHASHI ,  Makoto TERAO ,  Soma SHIRAISHI ,  Takami SATO ,  Yu NABETO ,  Ryosuke SAKAI

IPC: G06V10/20 ,  G06V10/25 ,  G06V10/70 ,  G06V20/40 ,  G06V30/224 ,  G06V40/10 ,  G06T7/70 ,  G06N5/04

Abstract: An image acquisition unit 110 acquires a plurality of images. The plurality of images include an object to be inferred. An image cut-out unit 120 cuts out an object region including the object from each of the plurality of images acquired by the image acquisition unit 110. An importance generation unit 130 generates importance information by processing the object region cut out by the image cut-out unit 120. The importance information indicates the importance of the object region when an object inference model is generated, and is generated for each object region, that is, for each image acquired by the image acquisition unit 110. A learning data generation unit 140 stores a plurality of object regions cut out by the image cut-out unit 120 and a plurality of pieces of importance information generated by the importance generation unit 130 in a learning data storage unit 150 as at least a part of the learning data.

公开(公告)号：US20230186591A1

公开(公告)日：2023-06-15

申请号：US17916823

申请日：2020-04-13

Applicant: NEC Corporation

Inventor： Ryosuke SAKAI

IPC: G06V10/44 ,  G06T7/68 ,  G06T7/73 ,  G06V20/70

CPC classification number: G06V10/44 ,  G06T7/68 ,  G06T7/73 ,  G06V20/70

Abstract: An information processing device 4 includes a feature point acquiring means 41A and a label determining means 43A. The feature point acquiring means 41A is configured to acquire, based on a captured image “Im” captured by an imaging unit 15A, a combination of positions of feature points of a symmetrical object with a symmetry and second labels defined to integrate or change first labels that are unique labels of the feature points based on the symmetry of the symmetrical object. The label determination means 43A is configured to determine a first label to be assigned to each of the feature points based on additional information to break the symmetry and the second labels.