Summary of Adaptive Self-training Framework For Fine-grained Scene Graph Generation, by Kibum Kim et al.
Adaptive Self-training Framework for Fine-grained Scene Graph Generation
by Kibum Kim, Kanghoon Yoon, Yeonjun In, Jinyoung Moon, Donghyun Kim, Chanyoung Park
First submitted to arxiv on: 18 Jan 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 tackles a crucial issue in Scene Graph Generation (SGG) models: the long-tailed predicate distribution problem. Specifically, they propose a Self-Training framework for SGG (ST-SGG) that assigns pseudo-labels to unannotated triplets to alleviate this issue. The approach, called Class-specific Adaptive Thresholding with Momentum (CATM), is model-agnostic and can be applied to any existing SGG models. Additionally, the paper introduces a Graph Structure Learner (GSL) that improves performance when using MPNN-based SGG models. Experimental results demonstrate the effectiveness of ST-SGG in enhancing fine-grained predicate class performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps Scene Graph Generation (SGG) models by solving a big problem: too many rare cases. Imagine trying to teach a machine how to recognize scenes, but there are too many different things it has to learn about at once. This makes it hard for the machine to get good at recognizing certain types of scenes. To fix this, researchers came up with a new way to train SGG models using unannotated data. They also created a special technique called CATM that helps improve performance on tricky cases. |
Keywords
» Artificial intelligence » Self training
