Title:
DEEP LEARNING-BASED LIP READING METHOD AND APPARATUS, ELECTRONIC DEVICE, AND MEDIUM
Document Type and Number:
WIPO Patent Application WO/2020/252922
Kind Code:
A1
Abstract:
Provided are a deep learning-based lip reading method and apparatus, an electronic device, and a medium. The deep learning-based lip reading method comprises: when a lip reading instruction is received, obtaining a video to be read (S10); splitting said video to obtain at least one sub-video (S11); inputting the at least one sub-video into a pre-trained lip reading model to obtain at least one sub-result (S12); inputting the at least one reading result into a configuration input method model for conversion, and outputting at least one segment of converted characters (S13); and splicing the at least one segment of converted characters to obtain the reading result (S14). According to the deep learning-based lip reading method, the effect is more visual, intelligent decision making is achieved, labor costs are reduced, the consumed time is shortened, and the user experience is improved.
Inventors:
DONG HONGTAO (CN)
Application Number:
PCT/CN2019/103368
Publication Date:
December 24, 2020
Filing Date:
August 29, 2019
Export Citation:
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G06K9/62; G06K9/00; G06N3/08; G10L15/25
Foreign References:
CN107992812A | 2018-05-04 | |||
CN108537207A | 2018-09-14 | |||
CN109858412A | 2019-06-07 | |||
CN106250829A | 2016-12-21 | |||
CN108921032A | 2018-11-30 | |||
CN108831472A | 2018-11-16 | |||
CN109409195A | 2019-03-01 | |||
US4769845A | 1988-09-06 |
Attorney, Agent or Firm:
SHENZHEN SCIENBIZIP INTELLECTUAL PROPERTY AGENCY CO., LTD. (CN)
Download PDF: