Summary of Promerge: Prompt and Merge For Unsupervised Instance Segmentation, by Dylan Li and Gyungin Shin
ProMerge: Prompt and Merge for Unsupervised Instance Segmentation
by Dylan Li, Gyungin Shin
First submitted to arxiv on: 27 Sep 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes Prompt and Merge (ProMerge), a novel approach to unsupervised instance segmentation that leverages self-supervised visual features. ProMerge initially groups patches using these features and then applies strategic merging, aided by background-based mask pruning. This method not only achieves competitive results but also reduces inference time compared to state-of-the-art normalized-cut-based approaches. Furthermore, when used as pseudo-labels for training an object detector, ProMerge surpasses the current leading unsupervised model on various challenging instance segmentation benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ProMerge is a new way to split objects in pictures without needing labeled data. It uses special computer vision features that were learned without human help. The program starts by grouping small pieces of the picture together based on these features, then it combines those groups into individual objects. This approach works well and runs fast. When ProMerge’s masks are used as fake labels to train an object detector, the result is better than the current best unsupervised model for this task. |
Keywords
» Artificial intelligence » Inference » Instance segmentation » Mask » Prompt » Pruning » Self supervised » Unsupervised
