Summary of Hierarchical Multi-label Classification with Missing Information For Benthic Habitat Imagery, by Isaac Xu et al.
Hierarchical Multi-Label Classification with Missing Information for Benthic Habitat Imagery
by Isaac Xu, Benjamin Misiuk, Scott C. Lowe, Martin Gillis, Craig J. Brown, Thomas Trappenberg
First submitted to arxiv on: 10 Sep 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 paper applies cutting-edge self-supervised learning methods to a large dataset of seafloor imagery, BenthicNet, to tackle a complex hierarchical multi-label (HML) classification task. The authors demonstrate the ability to conduct HML training in scenarios with missing annotation information, common when dealing with heterogeneous real-world data from multiple sources. They show that models pre-trained on larger in-domain benthic datasets outperform those trained on ImageNet, especially for local or regional scale projects. In the HML setting, self-supervised pre-training leads to deeper and more precise classifications. The authors hope this work establishes a benchmark for future models in automated underwater image annotation tasks and guides similar research in other domains. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses powerful new learning techniques on a big collection of pictures taken from the ocean floor. They test these methods by trying to categorize images with multiple levels of information, which is important when dealing with different types of data collected by various researchers. The results show that pre-training models on bigger datasets specific to the ocean floor leads to better performance compared to training them on more general data like ImageNet. This work aims to create a standard for future models and help others in this field. |
Keywords
» Artificial intelligence » Classification » Self supervised
