Few-Shot Learning in Semantic Segmentation

MPhil Thesis Defence


Title: "Few-Shot Learning in Semantic Segmentation"

By

Mr. Tianhan WEI


Abstract

Large-scale datasets such as ImageNet, PASCAL VOC, and COCO play important 
roles in the recent success of deep learning algorithms in image 
recognition tasks. However, there are not sufficient datasets specifically 
designed for few-shot learning, especially in the few-shot semantic 
segmentation domain. We build the first large-scale few-shot segmentation 
dataset, FSS-1000, which consists of 1000 object classes with pixelwise 
annotation of ground-truth segmentation. Unique in FSS-1000, our dataset 
contains a significant number of objects that have never been seen or 
annotated in previous datasets, such as tiny daily objects, merchandise, 
cartoon characters, logos, etc.

We adapt the structure of Relation Network to build our baseline few-shot 
segmentation model to validate FSS-1000. By adopting networks such as 
VGG-16, ResNet-101, and Inception as backbones, we found that training our 
model from scratch using FSS-1000 achieves competitive and even better 
results than training with weights pre-trained by ImageNet which is more 
than 100 times larger than FSS-1000. Both our approach and dataset are 
simple, effective, and extensible to learn the segmentation of new object 
classes given very few annotated training examples.


Date:  			Friday, 28 August 2020

Time:			3:00pm - 5:00pm

Zoom meeting:		https://hkust.zoom.us/j/92449193088

Committee Members:	Prof. Chi-Keung Tang (Supervisor)
 			Dr. Qifeng Chen (Chairperson)
 			Prof. Pedro Sander


**** ALL are Welcome ****