Title:
LEARNED MODEL, LEARNING DEVICE, LEARNING METHOD, AND LEARNING PROGRAM
Document Type and Number:
Japanese Patent JP2021033395
Kind Code:
A
Abstract:
To provide a learned model which, when an object having a part concealed is measured, allows the object to be recognized including the concealed part.SOLUTION: The learned model is generated through a step S12 of inputting measurement data to generate a feature map where feature quantities extracted from the measurement data are set in a coordinate space having correspondence relation with a measurement object space, steps S13 and S15 of calculating a contribution map defining a contribution of each of elements of the feature map to detection, and a step S18 of inputting the feature map and the contribution map and using the elements of the feature map with weights corresponding to contributions of the elements to derive prescribed information and updating a learning model.SELECTED DRAWING: Figure 5
Inventors:
NAKAMURA TOMOHIKO
Application Number:
JP2019149435A
Publication Date:
March 01, 2021
Filing Date:
August 16, 2019
Export Citation:
Assignee:
SECOM CO LTD
International Classes:
G06T7/00
Domestic Patent References:
| JP2011186633A | 2011-09-22 | |||
| JP2014203133A | 2014-10-27 |
Foreign References:
| WO2018173108A1 | 2018-09-27 |
Other References:
岡見 和樹: "空間再構成のための遮蔽に頑健な骨格推定技術に関する一検討", 電子情報通信学会技術研究報告, vol. 118, no. 266, JPN6023024918, 18 October 2018 (2018-10-18), JP, pages 65 - 69, ISSN: 0005086440
HUIYANG ZHANG: "Orientation and Occlusion Aware Multi-Person Pose Estimation using Multi-Task Deep Learning Network", 2019 IEEE INTERNATIONAL CONFERENCE ON CONSUMER ELECTRONICS (ICCE), JPN6023024917, 13 January 2019 (2019-01-13), US, pages 5, ISSN: 0005086441
HUIYANG ZHANG: "Orientation and Occlusion Aware Multi-Person Pose Estimation using Multi-Task Deep Learning Network", 2019 IEEE INTERNATIONAL CONFERENCE ON CONSUMER ELECTRONICS (ICCE), JPN6023024917, 13 January 2019 (2019-01-13), US, pages 5, ISSN: 0005086441
Attorney, Agent or Firm:
Haruka International Patent Office