Title:
GENERATING IMAGES WITH SMALL OBJECTS FOR TRAINING A PRUNED SUPER-RESOLUTION NETWORK
Document Type and Number:
WIPO Patent Application WO/2024/072021
Kind Code:
A1
Abstract:
An embodiment provides a method comprising detecting at least one object displayed within at least one input frame of an input video. The method further comprises cropping, from the at least one input frame, at least one cropped image including the at least one object. The method further comprises generating at least one training image by overlaying simulated text on the at least one cropped image. The method further comprises providing the at least one training image to a pruned convolutional neural network (CNN). The pruned CNN learns, from the at least one training image, to reconstruct objects and textual regions during image super-resolution.
Inventors:
JNAWALI KAMAL (US)
BAU TIEN CHENG (US)
KIM JOONSOO (US)
BAU TIEN CHENG (US)
KIM JOONSOO (US)
Application Number:
PCT/KR2023/014871
Publication Date:
April 04, 2024
Filing Date:
September 26, 2023
Export Citation:
Assignee:
SAMSUNG ELECTRONICS CO LTD (KR)
International Classes:
G06T3/40; G06N3/0464; G06N3/047; G06N3/08; G06N7/01
Domestic Patent References:
| WO2021225296A1 | 2021-11-11 | |||
| WO2022179492A1 | 2022-09-01 | |||
| WO2022037642A1 | 2022-02-24 |
Foreign References:
| US10970645B1 | 2021-04-06 |
Other References:
ZEJIANG HOU ET AL.: "Efficient Image Super Resolution Via Channel Discriminative Deep Neural Network Pruning", 2020 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP, 4 May 2020 (2020-05-04), pages 3647 - 3651, XP033793673, DOI: 10.1109/ICASSP40776.2020.9054019
Attorney, Agent or Firm:
Y.P.LEE, MOCK & PARTNERS (KR)
Download PDF: