EvoLogics Builds Subsea Perception Models with FiftyOne

Introduction: EvoLogics

EvoLogics designs and manufactures underwater communication systems,
positioning networks, and autonomous vehicles for the world's most demanding subsea environments. Based in Germany, the company has built technology for clients across commercial, offshore, defense, and research sectors.

Operating at depths and in conditions unreachable by conventional means, EvoLogics autonomous vehicles locate survivors in search and rescue operations, inspect subsea pipelines and infrastructure, monitor ocean environments, and support naval mine countermeasure activities. Their systems process multimodal data e.g. sonar data from various types, underwater and surface-mounted cameras in real time, during live missions.

“FiftyOne has become an important element of our dataset and model development pipeline. Since incorporating the tool into our workflows, we gained a better overview over what data is the best to be used to train and evaluate our models. As a result, we can deliver continuous dataset optimizations into production faster, as our model training pipelines read data directly from FiftyOne” — Dmytro Prylipko, ML Engineer, EvoLogics

Unique model development challenges for the underwater environment

Building perception models for underwater environments comes with unique and highly specialized operating conditions. Since sonar imagery is uniquely domain-specific, off-the-shelf pre-training models do not apply. Additionally, the same object can appear differently across missions depending on the sonar's scanning angle and pass. To capture these dynamics, annotation guidelines - such as targeting criteria and class assignments - are refined via a highly agile, data-driven feedback loop between annotation and performance analysis.

Before deploying FiftyOne, EvoLogics handled its data and model workflows with proprietary in-house solutions. This required dedicated development and maintenance.

Why FiftyOne: A fully programmatic SDK and on-prem deployment

When choosing a tool for EvoLogics' specialized use case, two requirements were non-negotiable: a full programmatic SDK to seamlessly integrate data processing into EvoLogics’ automated deployment workflows, and an on-premise deployment to keep sensitive mission data secure and off the public cloud.

FiftyOne met both with its SDK and GUI approach to development and deployment flexibility.

"FiftyOne gives us the ability to quickly observe and explore data in different ways - finding outliers and ensuring data consistency across the datasets we possess. And we had the requirement to be able to do it from a script, not just a UI." -Dmytro Prylipko, ML Engineer, EvoLogics

How EvoLogics uses FiftyOne

The Evologics team collects data from various sources such as underwater and surface-mounted cameras, sonars e.g., for various purposes like underwater object detection and autonomous navigation. Automated workflows process and standardize complex subsea sensor data, managing them efficiently within FiftyOne.

Dataset search and query. Since a single sidescan dataset covers multiple use cases, the team uses FiftyOne dataset views and filtering capabilities to slice the right subset for each model rather than duplicating data across separate training sets. This keeps the source of truth clean while giving every project the selection it needs.

Dataset optimization. Using FiftyOne Brain enabled the team to structure and cluster extensive subsea datasets more efficiently by visual appearance. Furthermore, leveraging EvoLogics' highly specialized pre-trained models, samples were pre-annotated to minimize human effort while ensuring maximum training readiness.

Visual QA and Dataset Refinement. Engineers inspect data distributions, run queries against metadata, and verify data quality before they reach training pipelines. When a model flags a false positive, the team loads predictions back into FiftyOne to refine and optimize model accuracy based on ground-truth data.

Interactive model evaluation. Teams load model predictions directly into
FiftyOne, utilizing advanced workflows to perform granular class-level analysis. This enables the team to rigorously validate edge cases to maximize model accuracy and optimize dataset coverage.

Quick navigation to events of interest. FiftyOne automatically segments extensive mission videos into concise clips of interest. This accelerates high-value data analysis by eliminating the need to search through full mission recordings.

Results: Enhanced data visibility accelerates how EvoLogics build highly accurate perception models

Since using the tool in their complex pipeline, Evologics now seamlessly
orchestrates and optimizes their data and model workflows at scale. FiftyOne
has also expanded what the team can do with their data. For classification
projects initiated dynamically with targeted seed labels, the team uses FiftyOne's embedding extractors and visualization capabilities to build out datasets faster using semi-supervised workflows.

Multimodal workflows

Centralizing complex, multimodal data pipelines drives operational efficiency. FiftyOne Enterprise's grouped datasets feature delivers a unified interface to seamlessly visualize, query, annotate, and evaluate across disparate data modalities at scale.

Learn about FiftyOne’s multimodal support →

How EvoLogics Powers the World’s Most Challenging Subsea Missions with FiftyOne