Summary of Fine-grained Scene Graph Generation Via Sample-level Bias Prediction, by Yansheng Li et al.
Fine-Grained Scene Graph Generation via Sample-Level Bias Prediction
by Yansheng Li, Tingzhu Wang, Kang Wu, Linlin Wang, Xin Guo, Wenbin Wang
First submitted to arxiv on: 27 Jul 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 a novel approach to Scene Graph Generation (SGG), which aims to better serve downstream tasks by exploring relationships between objects in images and generating scene summary graphs. However, the long-tailed problem has limited the quality of scene graphs, with coarse-grained relationships dominating over fine-grained ones. The authors introduce a Sample-Level Bias Prediction (SBP) method for fine-grained SGG (SBG), which predicts sample-specific biases to refine original relationship predictions. The approach consists of training a classic SGG model and constructing a correction bias set, followed by learning to predict these biases using a Bias-Oriented Generative Adversarial Network (BGAN). Experimental results on VG, GQA, and VG-1800 datasets demonstrate that SBG outperforms state-of-the-art methods in terms of Average@K for three mainstream SGG models. Specifically, SBG shows a significant average improvement of 5.6%, 3.9%, and 3.2% on Average@K for tasks PredCls, SGCls, and SGDet, respectively. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Scene graph generation aims to explore relationships between objects in images and generate scene summary graphs. However, the long-tailed problem has limited the quality of scene graphs, with coarse-grained relationships dominating over fine-grained ones. The authors introduce a new approach that predicts sample-specific biases to refine original relationship predictions. This approach uses a classic SGG model and a bias-oriented generative adversarial network (BGAN) to predict these biases. The results show that this approach outperforms state-of-the-art methods in terms of average@K. |
Keywords
» Artificial intelligence » Generative adversarial network
