Software Development for Underwater Computer Vision

AI-based software modules for the automated detection, classification, measurement and tracking of objects in underwater images and videos.

Fraunhofer IGD develops robust software modules for the automated analysis of underwater images and videos. AI-based methods detect, classify, measure and track objects and can be integrated into marine technology systems, inspection platforms and monitoring processes.

Robust Analysis of Underwater Imagery

Fraunhofer IGD develops software modules that automatically analyze underwater images and videos. The methods detect, classify, measure and track relevant objects. A particular focus is on delivering reliable results despite color distortion, turbidity and scattered light. This enables large volumes of data to be analyzed at scale while reducing the amount of manual work required from domain experts.

Dashboard for the automated analysis of underwater videos, showing camera footage from different locations, fish classifications and graphical evaluations.
© Fraunhofer IGD

Modular Integration from Edge to Cloud

The solutions are primarily based on machine learning and tailored to the specific application. The range of services extends from real-time processing on edge systems to cloud-based analysis. Training approaches designed for limited data availability and Fraunhofer IGD’s own GPU infrastructure also support the training of complex models. The software modules are developed to integrate seamlessly into existing customer systems, such as marine technology equipment, inspection platforms and monitoring processes. Users benefit from higher throughput, reproducible results and faster access to reliable information for monitoring, inspection and documentation.

Underwater image of an aquaculture facility showing several fish detected and measured by an AI system using colored bounding boxes.
© Fraunhofer IGD

Proven Applications in Fish Monitoring and Aquaculture

Fraunhofer IGD reference projects demonstrate the versatility of these capabilities. The “Sea Trout Analyzer & Annotator” automates the review and classification of extensive video footage. In the “OTC-ReefMonitor” project, Fraunhofer IGD is working with FIUM to develop algorithms for the optical detection and analysis of fish populations. “OTC-Marikultur” and “OTC-Mariculture2” use AI-based image and video analysis in aquaculture, including the analysis of fish behavior, health and stress indicators, as well as fish size measurement.

More about our research

 

Back to overview

Learn the essential facts about our team and its work.

 

Software Development for the Visualization of Maritime Sensor Data

 

Feasibility Studies for Maritime AI Applications