Unsupervised Instance Segmentation with Superpixels

2025/09/03 by Cuong Manh Hoang, Hoang, Cuong Manh
Engineering · #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Industrial Vision Systems and Defect Detection

paper · pdf · doi:10.48550/arxiv.2509.05352

openalex publication_date 2025/09/03 · openalex created_date 2025/10/11 · openalex updated_date 2026/07/28

Abstract

Instance segmentation is essential for numerous computer vision applications, including robotics, human-computer interaction, and autonomous driving. Currently, popular models bring impressive performance in instance segmentation by training with a large number of human annotations, which are costly to collect. For this reason, we present a new framework that efficiently and effectively segments objects without the need for human annotations. Firstly, a MultiCut algorithm is applied to self-supervised features for coarse mask segmentation. Then, a mask filter is employed to obtain high-quality coarse masks. To train the segmentation network, we compute a novel superpixel-guided mask loss, comprising hard loss and soft loss, with high-quality coarse masks and superpixels segmented from low-level image features. Lastly, a self-training process with a new adaptive loss is proposed to improve the quality of predicted masks. We conduct experiments on public datasets in instance segmentation and object detection to demonstrate the effectiveness of the proposed framework. The results show that the proposed framework outperforms previous state-of-the-art methods.

Citations

• unMORE: Unsupervised Multi-Object Segmentation via Center-Boundary Reasoning
• CutS3D: Cutting Semantics in 3D for 2D Unsupervised Instance Segmentation
• CuVLER: Enhanced Unsupervised Object Discoveries through Exhaustive Self-Supervised Transformers
• Unsupervised Universal Image Segmentation
• Cut and Learn for Unsupervised Object Detection and Instance Segmentation
• TokenCut: Segmenting Objects in Images and Videos with Self-supervised Transformer and Normalized Cut
• Discovering Object Masks with Transformers for Unsupervised Semantic Segmentation
• Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation
• Noisy Boundaries: Lemon or Lemonade for Semi-supervised Instance Segmentation?
• Unsupervised Semantic Segmentation by Distilling Feature Correspondences
• FreeSOLO: Learning to Segment Objects without Annotations
• Masked-attention Mask Transformer for Universal Image Segmentation
• RAMA: A Rapid Multicut Algorithm on GPU
• ST++: Make Self-training Work Better for Semi-supervised Semantic Segmentation
• Emerging Properties in Self-Supervised Vision Transformers
• Emerging Properties in Self-Supervised Vision Transformers
• Unidentified Video Objects: A Benchmark for Dense, Open-World Segmentation
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• SSAP: Single-Shot Instance Segmentation With Affinity Pyramid
• Learnable Tree Filter for Structure-preserving Feature Transform
• Attention Is All You Need
• Mask R-CNN
• Superpixel Segmentation Using Gaussian Mixture Model
• Deep Residual Learning for Image Recognition
• YFCC100M
• Microsoft COCO: Common Objects in Context
• Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
• Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

Related

• ProMerge: Prompt and Merge for Unsupervised Instance Segmentation
• Where are the Masks: Instance Segmentation with Image-level Supervision
• Weakly Supervised Instance Segmentation Using Hybrid Network
• Unsupervised Discovery of the Long-Tail in Instance Segmentation Using Hierarchical Self-Supervision
• Cut and Learn for Unsupervised Object Detection and Instance Segmentation