Towards Learning Object Detectors with Limited Data for Industrial Applications
Towards Learning Object Detectors with Limited Data for Industrial Applications
Karim Guirguis
English | 2024 | ISBN: 9783731513896 | 262 Pages | True PDF | 52 MB
Karim Guirguis
English | 2024 | ISBN: 9783731513896 | 262 Pages | True PDF | 52 MB
In this dissertation, three novel Generalized Few-Shot Object Detection (G-FSOD) approaches are presented to minimize the forgetting of previously learned classes while learning new classes with limited data. The first two approaches reduce the forgetting of base classes if they are still available during training. The third approach, for scenarios without base data, uses knowledge distillation to improve the knowledge transfer.