Summary of Graphkd: Exploring Knowledge Distillation Towards Document Object Detection with Structured Graph Creation, by Ayan Banerjee et al.
GraphKD: Exploring Knowledge Distillation Towards Document Object Detection with Structured Graph Creation
by Ayan Banerjee, Sanket Biswas, Josep Lladós, Umapada Pal
First submitted to arxiv on: 17 Feb 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 A graph-based knowledge distillation framework is proposed to improve object detection in documents. Large models can be computationally expensive and impractical for deployment on resource-constrained devices, so the framework aims to create small and efficient models that retain high accuracy. The method uses a structured graph with nodes containing proposal-level features and edges representing relationships between different regions. To reduce text bias, an adaptive node sampling strategy is designed to prioritize non-text nodes. The framework transfers knowledge from a teacher model to a student model through a distillation loss, effectively capturing local and global information. Experimental results on competitive benchmarks demonstrate that the proposed approach outperforms current state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to detect objects in documents is being developed. This method uses a special kind of graph to help small computers recognize important parts of documents quickly and accurately. The goal is to make this process work well even on devices with limited resources. To make sure the results are fair, the system tries to balance the importance of different parts of the document. This approach has been tested and shown to be better than other current methods. |
Keywords
* Artificial intelligence * Distillation * Knowledge distillation * Object detection * Student model * Teacher model
