Summary of A Fair Ranking and New Model For Panoptic Scene Graph Generation, by Julian Lorenz et al.
A Fair Ranking and New Model for Panoptic Scene Graph Generation
by Julian Lorenz, Alexander Pest, Daniel Kienzle, Katja Ludwig, Rainer Lienhart
First submitted to arxiv on: 12 Jul 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 addresses the issue of incorrect evaluation in panoptic scene graph generation (PSGG), where multiple masks for the same object can lead to inflated scores. The authors provide a fair ranking over existing PSGG models, correcting the previous flawed protocol. The results show that two-stage methods are competitive with one-stage methods, contrary to recent findings. Furthermore, the paper introduces the Decoupled SceneFormer (DSFormer), a novel two-stage model that outperforms all existing scene graph models by a large margin on the corrected evaluation. DSFormer encodes subject and object masks directly into feature space. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper fixes a problem with how we evaluate pictures to understand what’s going on in them. Right now, we’re giving too much credit to some methods because they can cheat by treating different parts of the same thing as separate things. The authors make it fair for all methods and find that two-stage approaches are just as good as one-stage ones. They also create a new way to understand pictures called Decoupled SceneFormer (DSFormer) that does better than other ways. |
