Remote sensing image automatic annotation method based on multi-model coupling evaluation

• Main Authors: RYU BYUNG-GI, ZHANG JUNSHAN, ZUO YULIN, REN MIXIN, TANG YUFENG, HAN YONG
• Format: Patent
• Language: Chinese; English
• Published: 28.05.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; PHYSICS

The invention discloses a remote sensing image automatic labeling method based on multi-model coupling evaluation, and the method comprises the following steps: S1, obtaining a remote sensing image data set, carrying out the manual fine labeling, and carrying out the counting of the number of labeled categories; s2, setting a category number threshold value, and expanding the sample number for categories with the category number smaller than the threshold value; s3, training is carried out to obtain respective optimal model weight parameters, and a trained target detection network model is obtained; the optimal accuracy is determined through the evaluation indexes; s4, a to-be-marked data set is sent into the obtained optimal model for prediction, and a label file is generated; and S5, screening, grouping and evaluating the generated label files, and outputting an original image and final label information. According to the method, the number of samples can be expanded before modeling training of the automati