University of Illinois Urbana-Champaign

Improving few-shot object detection by saving and hallucinating examples

Zhang, Weilin

Content Files
ZHANG-THESIS-2020.pdf

Permalink

https://hdl.handle.net/2142/109535

Description

Title
Improving few-shot object detection by saving and hallucinating examples
Author(s)
Zhang, Weilin
Issue Date
2020-12-09
Director of Research (if dissertation) or Advisor (if thesis)
Forsyth, David Alexander
Department of Study
Computer Science
Discipline
Computer Science
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
M.S.
Degree Level
Thesis
Date of Ingest
2021-03-05T21:42:52Z
Keyword(s)
  • Object Detection
  • Few-Shot Learning
Abstract
Learning to detect an object in an image from very few training examples - few-shot object detection - is challenging, because the classifier that sees proposal boxes has very little training data. A particularly challenging training regime occurs when there are one or two training examples. In this case, if the region proposal network (RPN) misses even one high intersection-over-union (IOU) training box, the classifier's model of how object appearance varies can be severely impacted. We use multiple distinct yet cooperating RPN's. Our RPN's are trained to be different, but not too different; doing so yields significant performance improvements over state of the art for COCO and PASCAL VOC in the very few-shot setting. This effect appears to be independent of the choice of classifier or dataset. However, under the very low-shot regime, even if all high IOU boxes are used to train the classifier, the variations are still insufficient to train the classifier in novel classes. We propose to build an even better model of variation in novel classes by transferring the shared within-class variation from base classes. We introduce a hallucinator network and insert it into a modern object detector model, which learns to generate additional training examples in the Region of Interest (ROI's) feature space. This approach yields further performance improvements on two state-of-the-art few-shot detectors with different proposal generation processes.
Graduation Semester
2020-12
Type of Resource
Thesis
Permalink
http://hdl.handle.net/2142/109535
Copyright and License Information
Copyright 2020 Weilin Zhang

Owning Collections

Graduate Dissertations and Theses at Illinois PRIMARY
Graduate Theses and Dissertations at Illinois
Dissertations and Theses - Computer Science
Dissertations and Theses from the Siebel School of Computer Science