Summary of Str-cert: Robustness Certification For Deep Text Recognition on Deep Learning Pipelines and Vision Transformers, by Daqian Shao et al.
STR-Cert: Robustness Certification for Deep Text Recognition on Deep Learning Pipelines and Vision Transformers
by Daqian Shao, Lukas Fesser, Marta Kwiatkowska
First submitted to arxiv on: 28 Nov 2023
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 The proposed method, STR-Cert, extends the DeepPoly polyhedral verification framework to formally certify scene text recognition (STR) models against adversarial inputs. This is achieved by deriving novel polyhedral bounds and algorithms for key STR model components. The certification process is applied to three types of STR model architectures, including standard pipelines and Vision Transformers, on six benchmark datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you’re trying to recognize words in pictures, like street signs or license plates. You want to make sure your machine learning model can handle tricky inputs that try to fool it. That’s what robustness certification is all about. Researchers have been working on this for neural networks, but now they’re tackling a harder problem: scene text recognition. This is when you try to read words in pictures. It’s really important because we use these models in self-driving cars and other safety-critical applications. |
Keywords
* Artificial intelligence * Machine learning
