Segmentation
Segmentation on World Data Ocean: a running collection of 3 stories we have gathered and hand-picked because they are worth your time. Every post here touches on segmentation in some way — the news, the analysis, the deep dives, and the occasional surprise find. A Global Hub for Ocean Intelligence 4 World Data Ocean is a centralized digital platform where researchers, scientists, and ocean enthusiasts converge to explore, analyze, and… New stories are added to this page as we find them, so check back if you want to keep up with what is happening around segmentation, or subscribe to the RSS feed to get them as soon as they are published. Browse the collection below, or head back to the homepage to see everything World Data Ocean is covering right now.
Ocean-aware deep learning for civilian maritime object detection and tracking in complex ocean environments: a comprehensive review
Reliable maritime object detection and tracking are paramount for civilian ocean engineering, supporting applications from vessel traffic monitoring to environmental risk assessment. However, dynamic ocean conditions—including sea clutter and limited data—present significant challenges. This review synthesizes recent advances in ocean-aware deep learning, examining techniques utilizing optical, radar, and other sensor data alongside AIS information. Addressing limitations in annotation and real-time deployment, future research should prioritize physics-informed learning and data fusion.
Improving cross-flight plastic litter segmentation with ConvNeXt V2 U-Net for UAV SWIR hyperspectral imagery
Accurate spatial monitoring of plastic litter is crucial for effective ecosystem management, and UAV short-wave infrared (SWIR) hyperspectral imagery offers high-resolution detection capabilities. However, cross-flight variability often limits operational utility. This study addresses this challenge by presenting a novel framework utilizing a compact ConvNeXt V2-based U-Net, achieving significant improvements in plastic litter segmentation across diverse acquisition conditions. Results demonstrate a substantial increase in Dice score (78.2%) and IoU (64.
PICOGRAM - informing coral reef resilience-based management through prediction of individual coral organismal growth, recruitment, and mortality
PICOGRAM introduces a novel, open-source framework for scalable coral reef monitoring, addressing the limitations of manual annotation in image-based assessments. Inspired by the NASA NeMO-Net ecosystem, PICOGRAM leverages a Segment Anything Model (SAM) to automate coral colony detection and percent-cover estimation from underwater imagery and orthomosaics. Evaluated against expert annotations, PICOGRAM achieves high accuracy (IoU 87.5% in-domain) and calibrated quality scores, enabling efficient refinement with minimal user input.