Semantic segmentation model training method, semantic segmentation method and device

The invention provides a training method of a semantic segmentation model and a semantic segmentation method and device, and relates to the technical field of image processing. The semantic segmentation model training method comprises the steps of obtaining sample image data, wherein the sample imag...

Full description

Saved in:
Bibliographic Details
Main Authors LIU WEI, HU JUN, LYU JIN, JIN SHU, ZHOU TING
Format Patent
LanguageChinese
English
Published 22.12.2023
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The invention provides a training method of a semantic segmentation model and a semantic segmentation method and device, and relates to the technical field of image processing. The semantic segmentation model training method comprises the steps of obtaining sample image data, wherein the sample image data comprises a first sample image configured with a label and a second sample image not configured with a label; based on the first sample image, knowledge distillation training processing is carried out on a student model by using a teacher model, the trained student model is determined as an initial semantic segmentation model, the teacher model is a complex model pre-trained based on the first sample image, and the initial semantic segmentation model is determined as a semantic segmentation model; the knowledge distillation training processing comprises feature layer distillation training and logic layer distillation training; and based on the second sample image, performing knowledge distillation training p
Bibliography:Application Number: CN202311237794