PNPDet : efficient few-shot detection without forgetting via Plug-and-Play sub-networks

• Main Authors: Zhang, Gongjie, Cui, Kaiwen, Wu, Rongliang, Lu, Shijian, Tian, Yonghong
• Other Authors: School of Electrical and Electronic Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2021
• Subjects: Engineering::Computer science and engineering; Few-Shot Object Detection; PNPDet
• Online Access: https://hdl.handle.net/10356/146204
• Institution: Nanyang Technological University

The human visual system can detect objects of unseen categories from merely a few examples. However, such capability remains absent in state-of-the-art detectors. To bridge this gap, several attempts have been proposed to perform few-shot detection by incorporating meta-learning techniques. Such met...