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

Abstract

Over the past few years, we have witnessed the success of deep learning in image recognition thanks to the availability of large-scale human-annotated datasets such as PASCAL VOC, ImageNet, and COCO. Although these datasets have covered a wide range of object categories, there are still a significant number of objects that are not included. Can we perform the same task without a lot of human annotations? In this paper, we are interested in few-shot object segmentation where the number of annotated training examples are limited to 5 only. To evaluate and validate the performance of our approach, we have built a 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 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 build our baseline model using standard backbone networks such as VGG-16, ResNet-101, and Inception. To our surprise, we found that training our model from scratch using FSS-1000 achieves comparable 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 easily extensible to learn segmentation of new object classes given very few annotated training examples. Dataset is available at https://github.com/HKUSTCV/FSS-1000.

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Prototypical Networks for Few-shot Learning
• Fully Convolutional Networks for Semantic Segmentation
• ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
• The Open Images Dataset V4: Unified image classification, object detection, and visual relationship detection at scale
• Learning feed-forward one-shot learners
• Few-Shot Segmentation Propagation with Guided Networks

Cited by in corpus (19)

• A Survey on Long-Tailed Visual Recognition
• Rethinking Space-Time Networks with Improved Memory Coverage for Efficient Video Object Segmentation
• Semi-supervised few-shot learning for medical image segmentation
• Hypercorrelation Squeeze for Few-Shot Segmentation
• On the Binding Problem in Artificial Neural Networks
• Prior Guided Feature Enrichment Network for Few-Shot Segmentation
• Semantic Communication with Conceptual Spaces
• One-Shot Object Detection without Fine-Tuning
• HODOR: High-level Object Descriptors for Object Re-segmentation in Video Learned from Static Images
• Few-Shot Segmentation with Global and Local Contrastive Learning
• Modular Interactive Video Object Segmentation: Interaction-to-Mask, Propagation and Difference-Aware Fusion
• PFENet++: Boosting Few-shot Semantic Segmentation with the Noise-filtered Context-aware Prior Mask
• Meta-Learning Initializations for Image Segmentation
• Attentional Prototype Inference for Few-Shot Segmentation
• CAFENet: Class-Agnostic Few-Shot Edge Detection Network
• Few-shot Medical Image Segmentation using a Global Correlation Network with Discriminative Embedding
• Bidirectional RNN-based Few Shot Learning for 3D Medical Image Segmentation
• Prototypical Region Proposal Networks for Few-Shot Localization and Classification
• Dataset Bias in Few-shot Image Recognition