Authors: Pengming Feng, Mingjie Xie, Hongning Liu, Xuanjia Zhao, Guangjun He, Xueliang Zhang, Jian Guan

Published on: February 06, 2024

Impact Score: 8.45

Arxiv code: Arxiv:2402.03708

Summary

• What is new: Introduction of the SISP dataset and the DFRInst method for fine-grained ship instance segmentation in satellite images.
• Why this is important: Existing datasets for ship instance segmentation lack fine-grained information, pixel-wise localization, image diversity, and variations.
• What the research proposes: Creation of the SISP dataset with 56,693 well-annotated ship instances across 10,000 images and introduction of the DFRInst network for improved segmentation.
• Results: The DFRInst method outperforms existing state-of-the-art methods in accurately segmenting ship instances on the SISP dataset.

Technical Details

Technological frameworks used: Dynamic Feature Refinement-assist Instance segmentation network (DFRInst)

Models used: Instance segmentation models

Data used: SISP dataset from SuperView-1 satellite images

Potential Impact

Satellite imaging companies, maritime monitoring services, and defense sectors

Want to implement this idea in a business?

We have generated a startup concept here: NavInsight.