Summary of Wake Vision: a Tailored Dataset and Benchmark Suite For Tinyml Computer Vision Applications, by Colby Banbury et al.
Wake Vision: A Tailored Dataset and Benchmark Suite for TinyML Computer Vision Applications
by Colby Banbury, Emil Njor, Andrea Mattia Garavagno, Matthew Stewart, Pete Warden, Manjunath Kudlur, Nat Jeffries, Xenofon Fafoutis, Vijay Janapa Reddi
First submitted to arxiv on: 1 May 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 The Wake Vision dataset is a large-scale collection of over 6 million images used to develop person detection models for low-power devices. This dataset consists of two variants: Wake Vision (Large) and Wake Vision (Quality), which are designed to improve model performance through pretraining, knowledge distillation, and manual labeling. The manually labeled validation and test sets achieve error rates as low as 2.2%, outperforming previous standards. Additionally, five benchmark sets are introduced to evaluate model performance in real-world scenarios, considering varying lighting, camera distances, and demographic characteristics. Training with Wake Vision improves accuracy by 1.93% compared to existing datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Wake Vision is a big dataset for machines that can see! It has over 6 million pictures of people, which helps computer models learn to detect people better. The dataset is special because it has two parts: one has lots of data, and the other has high-quality labels. This makes the model work better. They also created five tests to make sure the models are working well in real-life situations. Using this dataset makes the models 1.93% more accurate than before! You can use Wake Vision for free if you follow some rules. |
Keywords
» Artificial intelligence » Knowledge distillation » Pretraining
