NICEST: Noisy Label Correction and Training for Robust Scene Graph Generation

Nearly all existing scene graph generation (SGG) models have overlooked the ground-truth annotation qualities of mainstream SGG datasets, i.e., they assume: 1) all the manually annotated positive samples are equally correct; 2) all the un-annotated negative samples are absolutely background. In this...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	IEEE transactions on pattern analysis and machine intelligence Jg. 46; H. 10; S. 6873 - 6888
Hauptverfasser:	Li, Lin, Xiao, Jun, Shi, Hanrong, Zhang, Hanwang, Yang, Yi, Liu, Wei, Chen, Long
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	IEEE 01.10.2024
Schlagworte:	Annotations Benchmark testing Multi-Teacher knowledge distillation NIST Noise measurement noisy label learning out-of-distribution scene graph generation Task analysis Training Visualization
ISSN:	0162-8828, 2160-9292
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!