# Multi-granularity Interaction Simulation for Unsupervised Interactive Segmentation

Source: [Multi-granularity Interaction Simulation for Unsupervised Interactive Segmentation](<https://www.sota2.com/research/paper/68fa6ede1e209131cea42800>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T01:07:49.028Z

Authors: Kehan Li, Yian Zhao, Zhennan Wang, Zesen Cheng, Peng Jin, Xiangyang Ji, Li Yuan, Chang Liu, Jie Chen

Published: 2023-03-23T16:19:43.000Z

## Abstract

Interactive segmentation enables users to segment as needed by providing cues of objects, which introduces human-computer interaction for many fields, such as image editing and medical image analysis. Typically, massive and expansive pixel-level annotations are spent to train deep models by object-oriented interactions with manually labeled object masks. In this work, we reveal that informative interactions can be made by simulation with semantic-consistent yet diverse region exploration in an unsupervised paradigm. Concretely, we introduce a Multi-granularity Interaction Simulation (MIS) approach to open up a promising direction for unsupervised interactive segmentation. Drawing on the high-quality dense features produced by recent self-supervised models, we propose to gradually merge patches or regions with similar features to form more extensive regions and thus, every merged region serves as a semantic-meaningful multi-granularity proposal. By randomly sampling these proposals and simulating possible interactions based on them, we provide meaningful interaction at multiple granularities to teach the model to understand interactions. Our MIS significantly outperforms non-deep learning unsupervised methods and is even comparable with some previous deep-supervised methods without any annotation.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Interactive Segmentation on SBD](<https://www.sota2.com/research/sota/interactive-segmentation-on-sbd.md>): Interactive Segmentation; SBD; NoC @ 90% Target: 9.51.

- [Interactive Segmentation on GrabCut](<https://www.sota2.com/research/sota/interactive-segmentation-on-grabcut.md>): Interactive Segmentation; GrabCut; NoC@90: 2.32.

- [Interactive Segmentation on DAVIS](<https://www.sota2.com/research/sota/interactive-segmentation-on-davis.md>): Interactive Segmentation; DAVIS; NoC@90: 8.44.

- [Interactive Segmentation on Berkeley](<https://www.sota2.com/research/sota/interactive-segmentation-on-berkeley.md>): Interactive Segmentation; Berkeley; NoC@90: 4.58.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2303.13399>)

[PDF](<https://arxiv.org/pdf/2303.13399.pdf>)
