Summary of Gating Syn-to-real Knowledge For Pedestrian Crossing Prediction in Safe Driving, by Jie Bai et al.
Gating Syn-to-Real Knowledge for Pedestrian Crossing Prediction in Safe Driving
by Jie Bai, Jianwu Fang, Yisheng Lv, Chen Lv, Jianru Xue, Zhengguo Li
First submitted to arxiv on: 24 Aug 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 research paper proposes a novel approach for pedestrian crossing prediction (PCP) in driving scenes. The authors leverage synthetic data with flexible variation to boost prediction performance using domain adaptation frameworks. They introduce a Gated Syn-to-Real Knowledge transfer approach, which has two main aims: designing suitable domain adaptation methods and transferring knowledge for specific situations through gated knowledge fusion. The framework includes three domain adaptation methods (style transfer, distribution approximation, and knowledge distillation) that handle various types of information such as visual, semantic, depth, location, etc. A Learnable Gated Unit is employed to fuse suitable cross-domain knowledge, resulting in superior PCP performance compared to state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Pedestrian crossing prediction is crucial for ensuring the safe operation of intelligent vehicles. To improve prediction performance, researchers are using synthetic data with flexible variation and domain adaptation frameworks. This paper proposes a new approach that transfers knowledge from synthetic data to real-world scenarios. The authors design a framework that combines three different methods for adapting domain knowledge, including style transfer, distribution approximation, and knowledge distillation. They also introduce a Learnable Gated Unit that fuses suitable cross-domain knowledge. The results show that this approach outperforms current state-of-the-art methods. |
Keywords
» Artificial intelligence » Domain adaptation » Knowledge distillation » Style transfer » Synthetic data
