Harmonizing Base and Novel Classes: A Class-Contrastive Approach for Generalized Few-Shot Segmentation (2023-03-24T00:00:00.000000Z)

TL;DR

This work proposes a class contrastive loss and a class relationship loss to regulate prototype updates and encourage a large distance between prototypes from different classes, thus distinguishing the classes from each other while maintaining the performance of the base classes.

Abstract

Current methods for few-shot segmentation (FSSeg) have mainly focused on improving the performance of novel classes while neglecting the performance of base classes. To overcome this limitation, the task of generalized few-shot semantic segmentation (GFSSeg) has been introduced, aiming to predict segmentation masks for both base and novel classes. However, the current prototype-based methods do not explicitly consider the relationship between base and novel classes when updating prototypes, leading to a limited performance in identifying true categories. To address this challenge, we propose a class contrastive loss and a class relationship loss to regulate prototype updates and encourage a large distance between prototypes from different classes, thus distinguishing the classes from each other while maintaining the performance of the base classes. Our proposed approach achieves new state-of-the-art performance for the generalized few-shot segmentation task on PASCAL VOC and MS COCO datasets.

Authors

Yuming Fang

8 papers

Guosheng Lin

8 papers

Chuan-Sheng Foo

2 papers

TL;DR

Abstract

Authors

References64 items

CRCNet: Few-Shot Segmentation with Cross-Reference and Region–Global Conditional Networks

Self-Support Few-Shot Semantic Segmentation

MSANet: Multi-Similarity and Attention Guidance for Boosting Few-Shot Segmentation

Distilling Knowledge From Object Classification to Aesthetics Assessment

Beyond the Prototype: Divide-and-conquer Proxies for Few-shot Segmentation

Integrative Few-Shot Learning for Classification and Segmentation

Learning What Not to Segment: A New Perspective on Few-Shot Segmentation

APANet: Adaptive Prototypes Alignment Network for Few-Shot Semantic Segmentation

ABPNet: Adaptive Background Modeling for Generalized Few Shot Segmentation

Deep Reasoning Network for Few-shot Semantic Segmentation

Few-Shot Semantic Segmentation with Cyclic Memory Network

Pixel Contrastive-Consistent Semi-Supervised Semantic Segmentation

Few-Shot Segmentation With Optimal Transport Matching and Message Flow

Cross-Image Region Mining With Region Prototypical Network for Weakly Supervised Segmentation

Few-Shot Segmentation with Global and Local Contrastive Learning

Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformer

Learning Meta-class Memory for Few-Shot Semantic Segmentation

Self-Supervision

Scale-Aware Graph Neural Network for Few-Shot Semantic Segmentation

Anti-aliasing Semantic Reconstruction for Few-Shot Semantic Segmentation

Learning a Few-shot Embedding Model with Contrastive Learning

Few-Shot Video Object Detection

Adaptive Prototype Learning and Allocation for Few-Shot Segmentation

Hypercorrelation Squeeze for Few-Shot Segmenation

Guided Co-Segmentation Network for Fast Video Object Segmentation

Self-Guided and Cross-Guided Learning for Few-Shot Segmentation

Mining Latent Classes for Few-shot Segmentation

Beyond Max-Margin: Class Margin Equilibrium for Few-shot Object Detection

Exploring Cross-Image Pixel Contrast for Semantic Segmentation

Contrastive Learning for Label Efficient Semantic Segmentation

Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?

Dynamic Extension Nets for Few-shot Semantic Segmentation

Weakly Supervised Segmentation with Maximum Bipartite Graph Matching

Generalized Few-shot Semantic Segmentation

BriNet: Towards Bridging the Intra-class and Inter-class Gaps in One-Shot Segmentation

Prior Guided Feature Enrichment Network for Few-Shot Segmentation

Self-Supervision with Superpixels: Training Few-shot Medical Image Segmentation without Annotation

Part-aware Prototype Network for Few-shot Semantic Segmentation

SimPropNet: Improved Similarity Propagation for Few-shot Image Segmentation

Self-Supervised Tuning for Few-Shot Segmentation

CRNet: Cross-Reference Networks for Few-Shot Segmentation

Weakly Supervised Few-shot Object Segmentation using Co-Attention with Visual and Semantic Embeddings

Cross Attention Network for Few-shot Classification

PANet: Few-Shot Image Semantic Segmentation With Prototype Alignment

FSS-1000: A 1000-Class Dataset for Few-Shot Segmentation

Image Block Augmentation for One-Shot Learning

Image Deformation Meta-Networks for One-Shot Learning

Finding Task-Relevant Features for Few-Shot Learning by Category Traversal

Revisiting Local Descriptor Based Image-To-Class Measure for Few-Shot Learning

Panoptic Feature Pyramid Networks

Task Agnostic Meta-Learning for Few-Shot Learning

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Pyramid Scene Parsing Network

Optimization as a Model for Few-Shot Learning

Attention to Scale: Scale-Aware Semantic Image Segmentation

Fully convolutional networks for semantic segmentation

Microsoft COCO: Common Objects in Context

Interaction-Matrix Based Personalized Image Aesthetics Assessment

Few-Shot Semantic Segmentation with Democratic Attention Networks

Splitting Vs. Merging: Mining Object Regions with Discrepancy and Intersection Loss for Weakly Supervised Semantic Segmentation

Few-Shot Semantic Segmentation with Prototype Learning

Siamese Neural Networks for One-Shot Image Recognition

Hypercor - relation squeeze for few - shot segmentation Hypercor - relation squeeze for few - shot segmentation

International Journal of Computer Vision manuscript No. (will be inserted by the editor) The PASCAL Visual Object Classes (VOC) Challenge

Field of Study

Journal Information

Name

Page

Volume

Venue Information

Name

Type

URL

Alternate Names