Title:
DEVICE AND METHOD FOR RECOGNIZING ACTIVITY IN SPORTS VIDEO BY USING CGAM
Document Type and Number:
WIPO Patent Application WO/2023/113105
Kind Code:
A1
Abstract:
The present invention relates to a device and a method for recognizing an activity in a sports video by using a CGAM. The device for recognizing an activity in a sports video by using a CGAM according to the present invention comprises: an object feature extraction unit for extracting an object feature value by multiplication of a first feature value and a second feature value, wherein the first feature value is output by sequentially inputting, into multiple convolution blocks and spatial attention modules (CBAMs) each being disposed between the convolution blocks, frames, the importance of which is divided in a temporal attention module (TAM) in which a video to be analyzed is input in units of frames and the importance among the frames is divided, and the second feature value is output by the CGAM which generates an object representation by compressing different pieces of object information output from each of the spatial attention modules; and an activity feature extraction unit for sequentially inputting the extracted object feature value into a recurrent neural network (RNN) and a fully-connected (FC) layer, and classifying a final activity from a probability value for each activity, estimated using a sigmoid function.
Inventors:
LEE SOOWON (KR)
RYU KWANGHYUN (KR)
RYU KWANGHYUN (KR)
Application Number:
PCT/KR2022/003141
Publication Date:
June 22, 2023
Filing Date:
March 07, 2022
Export Citation:
Assignee:
FOUNDATION SOONGSIL UNIV INDUSTRY COOPERATION (KR)
International Classes:
G06V40/20; G06V10/44; G06V10/62; G06V10/764; G06V20/40
Foreign References:
| KR20210026664A | 2021-03-10 | |||
| KR20210066694A | 2021-06-07 | |||
| KR20200106526A | 2020-09-14 |
Other References:
MINHAS RABIA A., JAVED ALI, IRTAZA AUN, MAHMOOD MUHAMMAD TARIQ, JOO YOUNG BOK: "Shot Classification of Field Sports Videos Using AlexNet Convolutional Neural Network", APPLIED SCIENCES, vol. 9, no. 3, pages 483, XP093072736, DOI: 10.3390/app9030483
CHEN CHUN-FU RICHARD; FAN QUANFU; PANDA RAMESWAR: "CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classification", 2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), IEEE, 10 October 2021 (2021-10-10), pages 347 - 356, XP034092544, DOI: 10.1109/ICCV48922.2021.00041
CHEN CHUN-FU RICHARD; FAN QUANFU; PANDA RAMESWAR: "CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classification", 2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), IEEE, 10 October 2021 (2021-10-10), pages 347 - 356, XP034092544, DOI: 10.1109/ICCV48922.2021.00041
Attorney, Agent or Firm:
YUN, Kuisang (KR)
Download PDF:
Previous Patent: PRESSURE SENSOR UNIT
Next Patent: AUTONOMOUS DRIVING ROBOT, CLOUD APPARATUS, AND LOCATION CORRECTION METHOD
Next Patent: AUTONOMOUS DRIVING ROBOT, CLOUD APPARATUS, AND LOCATION CORRECTION METHOD