United States

Accurate underwater object recognition remains challenging due to strong backscatter from the target, environment, and water column, as well as the limited use of polarization in conventional imaging systems. Most existing RGB cameras and LiDARs employ single or dual linear polarizations, leaving the rich interaction between light polarization and submerged objects underexploited. In this work, we present an multi-polarization LiDAR imaging and recognition pipeline that combines novel hardware with turbidity-aware computation. On the hardware side, we develop a metasurface-based LiDAR capable of transmitting and receiving four polarization states spanning linear and circular bases, yielding 16 transmit–receive polarization pair images per acquisition. On the algorithmic side, we introduce a turbidity-aware subset selection (TASS) framework that first performs turbidity-aware normalization (TAN), estimating a scene turbidity index from Stokes statistics (e.g., degree-of-polarization decay and return variance) and reweighting each polarization pair to equalize effective SNR and stabilize polarization structure. Lightweight scores are then computed that jointly capture contrast-to-backscatter ratio, polarization preservation (DoLP/angle-of-polarization stability), and inter-pair diversity (low mutual information). A simple greedy policy selects a budgeted subset of k polarization pairs, which are fused by a compact CNN for recognition. We evaluate our system on a submerged dataset of 10 alphanumeric targets imaged under three discrete turbidity levels and varied geometric configurations. Experiments show that (i) multi-polarization fusion significantly outperforms single-polarization baselines, (ii) models trained on optimized polarization subsets can surpass using all 16 pairs, and (iii) TASS with TAN improves accuracy over naive selection and matches or approaches the performance of full-pair fusion, while being more compute- and acquisition-efficient. Overall, our results demonstrate a practical, budget-conscious path toward turbidity-robust underwater object recognition without modifying existing multi-polarization LiDAR hardware.

NAML 2026

Metasurface-Based Polarimetric LiDAR and Turbidity-Aware Fusion for Underwater Imaging

polarization fusion.

stokes imaging

turbidityaware normalization

singlephoton detection

metasurface optics

polarimetric lidar

underwater imaging

subset selection

### Welcome to the 9th annual workshop on Naval Applications of Machine Learning!

NAML is the premier annual event showcasing current machine learning research relevant to Naval applications. We welcome participation from the defense community, industry, and academia. The NAML workshop is an opportunity for researchers from Navy research and development centers and our partner organizations to exchange technical ideas and to come together as a team. 
 Naval Information Warfare Center Pacific is partnering with AFCEA San Diego chapter to host the workshop.

The goals of NAML are:

* To spread awareness of current machine learning research relevant to Naval applications
* To connect machine learning researchers with experts in military requirements
* To build and strengthen collaborations within the defense research community

#### *"What we do as a Navy is a human business. Don’t pass up an opportunity to come together as a team. Don’t pass up leadership opportunities, and don’t outsource them to a PowerPoint. If you have an opportunity to come together as a team, to increase belonging, do it."*
Admiral John M. Richardson, 31st Chief of Naval Operations

Please log in to explore this event. You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Please log in to explore this event.

Registration for this workshop has now closed. We appreciate your interest and hope to welcome you at a future event. If you’d like to hear about similar upcoming events, please check our website or join our mailing list.

Registration for this workshop has now closed.

NAML 2025

The 9th annual workshop on Naval Applications of Machine Learning is an opportunity for researchers from Navy research and development centers and our partner organizations to exchange technical ideas and to come together as a team. The goals of NAML are to spread awareness of current machine learning research relevant to Naval applications, to connect machine learning researchers with experts in military requirements and to build and strengthen collaborations within the defense research community.

technical paper

### Welcome to the 10th annual workshop on Naval Applications of Machine Learning!

NAML is the premier annual event showcasing current machine learning research relevant to Naval applications. We welcome participation from the defense community, industry, and academia. The NAML workshop is an opportunity for researchers from Navy research and development centers and our partner organizations to exchange technical ideas and to come together as a team. 
Naval Information Warfare Center Pacific is partnering with AFCEA San Diego chapter to host the workshop.
#### On-Demand Access

* This is an on-demand site and all the recorded content from the workshop is available here. 
* Access to the recordings is exclusively available to registered participants.
* Each registrant has received an email notification, with instructions on how to access the content.

[![](https://assets.underline.io/markdown_image/1/image/60cd12aa376c191d3467ea2ee60944ea.png)](https://naml2026.org/)

The goals of NAML are:

* To spread awareness of current machine learning research relevant to Naval applications
* To connect machine learning researchers with experts in military requirements
* To build and strengthen collaborations within the defense research community

#### *"What we do as a Navy is a human business. Don’t pass up an opportunity to come together as a team. Don’t pass up leadership opportunities, and don’t outsource them to a PowerPoint. If you have an opportunity to come together as a team, to increase belonging, do it."*
Admiral John M. Richardson, 31st Chief of Naval Operations

This is an on-demand site. All recorded content from NAML 2026 is available here and access to it is exclusively available to registered participants.

**Please log in using the email address you registered with.** If you have any issues logging in, please contact us at <naml2026@underline.io>.

Let’s celebrate the past decade by investing in the next. 🚀

This is an on-demand site. All recorded content from NAML 2026 is available here and access to it is exclusively available to registered participants.

**Explore the latest advancements in machine learning tailored to Naval applications 🔎**

With on-demand access to NAML 2026, you can watch all lectures & browse poster presentations at your convenience, while also engaging directly with presenters via the Underline platform - ask questions, exchange ideas, and stay connected.

**If you still haven’t registered, you can do so now.**
👇
**Take advantage of a special one-time offer: full access for just $99, with complimentary registration for students, government, and military employees.**

**Get access here: https://naml2026-ondemand.paperform.co/**

Please check your spam folder, or contact our support team at **<naml2026@underline.io>** if you registered, but still haven't received an access email.

Let’s celebrate the past decade by investing in the next. 🚀

Please register!

The 10th Annual Workshop on Naval Applications of Machine Learning (NAML 2026) was held in San Diego, California, March 2-5, 2026. NAML is a forum for interdisciplinary experts from government, industry, and academia to discuss results of ongoing research on theory and applications of machine learning to address some of the most challenging problems faced by our naval forces. 



The Naval Innovative Science & Engineering (NISE) project, Localization of Aircraft using Several Shipboard Optical Sensors (LASSOS) addresses the critical need for autonomy and perception in degraded environments to maintain our fleet’s competitive advantage. LASSOS aims to estimate the position of incoming aircraft relative to ships using only Pan-Tilt-Zoom (PTZ) cameras, supporting both flight deck tracking to reduce sailor cognitive load and enable Unmanned Aerial Vehicle (UAV) recovery operations. While current commercial off-the-shelf solutions rely on active sensors like Light Detection and Ranging (LiDAR), Navy operations require Emissions Control (EMCON)-compatible passive sensing solutions.
Our technical approach comprises three key tasks: (1) developing machine learning algorithms for PTZ camera calibration, (2) implementing aircraft detection and tracking using object detection models to maintain aircraft within camera field of view, and (3) combining this information to achieve accurate 3D aircraft localization relative to the ship.
The first task, PTZ camera calibration, has two components: an offline calibration algorithm, and an online calibration algorithm. In the offline algorithm, a dataset is captured of several images from each camera containing a set of calibration patterns at different pan, tilt, and zoom values. This is used to estimate and optimize the intrinsic and extrinsic calibration parameters for each camera. The online algorithm is then used to estimate the calibration parameters in real time based on each camera’s current pan, tilt, and zoom value. The second task tracks aircraft using a YOLO-family object detection model. Once an aircraft is identified, the cameras automatically center the aircraft within its view to maintain a high-quality track. The third task takes the estimated calibration parameters from the first task, and the bounding boxes from the second task to provide a 3D position estimate of the aircraft.
This work directly addresses a critical capability gap for NAVAIR, as several UAV recovery systems currently under investigation require precise positional data of incoming aircraft. By providing this data through passive means, LASSOS enables these recovery systems to function in EMCON environments. Beyond recovery operations, the passive sensing technologies developed through LASSOS can be applied to numerous naval perception challenges, advancing autonomous systems across the fleet.

Localization of Aircraft using Several Shipboard Optical Sensors (LASSOS

Real-time perception in aerial and maritime domains is difficult because detection-plus-tracking pipelines depend on running detectors at every frame, making them slow and computationally expensive. To address this, we develop and present a suite of data-propagation strategies that allow detections to be reused or predicted across intermediate frames, enabling stronger detectors to run less frequently while maintaining accuracy and recall. Data propagation works by using previously computed detections and either copying them forward or updating them through geometric or appearance-based cues, effectively reducing detector load and increasing FPS. We study three cases: (1) Copy Strategy, which directly propagates prior detections; (2) Motion Strategy, which updates detections using a Kalman filter to estimate future positions; and (3) Copy-or-Motion with Re-ID, which selects the better propagation path based on feature similarity from a re-identification model. We evaluate these techniques on challenging aerial and maritime videos featuring diverse object sizes, motion blur, and dynamic backgrounds.
Our results show that no single strategy dominates across all conditions. Copy propagation struggles with small or fast objects but performs well for medium to large targets in slow-moving defense scenarios such as ships or aircraft. Motion propagation is more resilient to rapid movement but can drift with irregular motion patterns. Surprisingly, the combined Re-ID–based approach does not consistently outperform the simpler methods; reliance on feature matching increases latency and reduces the geometric advantages of motion modeling. Through qualitative and quantitative comparisons, we demonstrate how data propagation significantly boosts FPS while preserving detector quality. We also provide practical insights on selecting the appropriate propagation strategy based on object scale, movement characteristics, and application requirements, offering a foundation for building efficient real-time perception pipelines.


Data Propagation Techniques for Real-Time Tracking in Aerial and Maritime Scenarios

Recently, spatial computing and novel view rendering have gained significant traction with organizations increasingly investing R&D efforts in these technologies to develop applications for simulation, visualization and navigation. As part of the Intelligence Advanced Research Projects Activity’s (IARPA) WRIVA program, Accenture has been developing advanced novel view synthesis methodologies to tackle 3D rendering and site modeling at speed and at scale using incidental imagery to support mission planning and preparation, remote asset monitoring and inspection, visualization and virtual reality.

One of the challenges of current methodologies is their scalability and computational needs. Structure from motion pipelines contain spatial matching that calculates the proximity of all cameras using GPS information to generate a list of possible spatial nearest neighbors. It then proceeds to use salient image features and Nearest Neighbor Ratio Test Filtering to find inlier matches. This process scales quadratically with the number of input images, making the ingestion of thousands of input images computationally infeasible. To address the complex task of creating large reconstructions in a timely manner, we use an improved pipeline that addresses the bottlenecks present in feature extraction, feature matching, and reconstruction. 

For a scene with thousands of images and GPS data, a more robust approach would use hierarchical localization with a better feature extractor to account for large perspective changes and appearance changes. In addition, a better matcher would use better descriptors and context aware attention mechanisms. However, even with better features and matches, exhaustive image matching is not feasible for practical applications. Therefore, we first create a list of coarse matches using GPS data and then match robustly. With noisy GPS data, we assume up to a maximum of 20 meters error, with an average of 3-6 meters. However, the possible spatial neighbors per image remains vast. As a result, we use visual place recognition (VPR) to create a list of coarse matches. VPR models can run very quickly and account for time variations within images. We limit the number of matches to 25 per image by generating similarity scores between all the global descriptors of input images. Afterward, we perform geometric verification. Lastly, we use reconstruction methods that estimate the camera pose and structure for all images simultaneously and perform bundle adjustment in scalable steps.

In this talk we describe how we are researching state-of-the-art capabilities in 3D reconstruction and site modeling (e.g., Gaussian Splatting) to address scale. We will discuss how we are using available image metadata, camera pose estimation and diffusion models to build 3D site representations and reconstruct scenes given large number of input images from unconstrained sensor conditions.

BELLA - A Framework for Large Scale Novel View Rendering

High-Performance Computing (HPC) systems generate massive numerical simulation data that domain experts struggle to analyze efficiently, while Large Language Models (LLMs) excel at natural language but cannot interpret raw numerical outputs. We present a novel, domain-agnostic multimodal pipeline that bridges this gap by automatically transforming numerical simulation data into natural language, without requiring manual annotation, visualization, or expert labeling.

Our proposed pipeline includes a multi-modal model, obtained by fusing a state-of-the-art LLM (such as one from the Mistral or Llama families) with an encoding pipeline, which projects a numerical sample onto a meaningful latent space. The two components are fused by means of a cross-attention layer, which connects the textual and numerical information.

An integral part of the research is the training methodology, which includes the generation of representative data samples by means of numerical analysis tools. The data samples aim at linking the latent representation of the numerical data to quantities of interest for scientists, and their natural language descriptions.

As a proof of concept, we present an ongoing project, which is part of Destination Earth, European Commission’s flagship initiative aimed at creating an accurate digital twin of the Earth system. The project is a weather specialization of our suggested approach, fusing an existing AI model (AIFS, developed by ECMWF) with an open source LLM (from the Mistral family), to enable real-time analysis of numerical weather prediction data.

As this project is still in progress, we will highlight its achievements and the challenges we are currently facing.


Designing a multimodal LLM to analyze numerical data

Overview:
Capturing, analyzing, and exploiting lessons from operations allows improvement in NATO technology, tactics, and procedures. In order to utilize AI in the analysis of Lessons Learned, we must be able to recognize positive and negative results. To move towards these capabilities, experiments were performed on tracing the impact of lessons in subsequent doctrine. This paper discusses those experiments within the context of the RTG goals. Working with various versions of doctrine as well as the observations of operational staff during exercises, these experiments attempted to recognize when lessons are reflected in newer versions of doctrine. The development of Large Language Models (LLM) has shown that one of key factors in utilizing AI models is the ability to evaluate the results from a model not only to judge the results, but to improve the models. These experiments provide one piece of the evaluation required to apply AI to the LL process. 
Implementation:
Separating useful feedback from a large repository of Lessons Learned is challenging and time-consuming for NATO analysts. This project experimented with the ways LLMs can bring clarity and focus to their work. 
After cleaning and parsing the data, comparisons of cosine semantic similarity were performed in order to demonstrate the value of using visualizations to present this data. Through heat maps, bipartite graphs, and U-MAP, we were able to show how to simplify the analysis process using these automated processes. 
Additionally, we created knowledge graphs to explore the Lessons Learned repository. Using MemgraphLab and Python scripts, we are able to extract entities and properties from the Lessons Learned and use them to create a knowledge graph that is easy to query and use to quickly see patterns.
We envision future work to consist of creating a People Knowledge Graph that will use Natural Language Processing (NLP) techniques to extract skills from a Subject Matter Expert database and to connect these to Lessons Learned. This will facilitate interdisciplinary discussion and collaboration. 
Summary: 
The current presentation consists of our early exploratory work with this rich and multi-layered Lessons Learned database. We present our initial ideas, results, as well as our future plans. 


Tracing the Contributions of Lessons Learned

Infrared (IR) imaging systems, which are designed to detect heat radiating from objects in a scene, can capture features that may not otherwise appear in visible light systems. In a stratified ocean, objects and animals coming to the water surface can generate hydrodynamic trails due to an upwelling effect. These trails are detectable in the infrared and can persist for several minutes on the surface. We used a thermal dataset containing surfacing North Atlantic Right Whales (NARWs), collected as part of a prior health survey of the species, as targets of opportunity to train machine learning models to automatically detect the presence of whales and locate their hydrodynamic trails. Here, we review our model training methodologies and vision for the direction of this work. We are seeking to continue this research to develop a deep learning-based thermal imagery tool that, if integrated onto an aerial platform, could be deployed as an early-detection system to help reduce the risk of collisions and other harmful interactions with marine mammals.

A Deep Learning-Based Tool for the Automated Detection of Objects in Thermal Video Data

The “Dead Reckoning via Adaptive Parameter Estimation” project proposes a new approach for improving the positional accuracy of dead-reckoning by dynamically adapting parameters used for Kalman filter state estimation. Whereas typical methods use static estimates for these parameters, this new method uses measurements from multiple sensor sources (inertial, odometric, etc.) to model system operation and machine learning to adaptively adjust parameters. This data-driven approach improves position prediction and is particularly useful for systems reliant on inertial sensors (IMU) for navigating without GPS (e.g., unmanned vehicles) where accurate position estimates are required.

Dead Reckoning via Adaptive Parameter Estimation

NAUTICAL will establish a workforce development strategy, a competency development model, and a series of courses to build upon the professional cadre scientists, engineers, and technicians able to develop and advance artificial intelligence (AI) at NIWC Pacific and other warfare. The demand for an effective, capable AI workforce is greater than ever and will continue to grow because the world appears to be at the cusp of an AI revolution.  This project will focus on providing students with basic AI algorithmic theory and more practical AI implementation skills from a DoD perspective while limiting the mathematical aspects of AI. 
 
NAUTICAL will provide different sessions for basic, intermediate and advanced levels of AI coursework, developed primarily by NIWC AI SMEs for the Navy and DoD.  Basic AI for DoD 101 training will be conducted in multiple modules over the course of one to two days consisting of a series of lessons and programming assignments to cover various fundamental topics in the world of AI. Intermediate and advanced focus courses in specific stand-alone AI technology area will be provided in modules following the novice level lectures/exercises. Courses will be held virtually via Flankspeed Teams to provide flexibility and recordability for posterity.  
 
NAUTICAL will publicize our workforce development effort throughout NIWC Pacific and other Naval warfare centers. All DoD military, civilian, and contractors are welcome to partake in NAUTICAL.   
 
Guest lecturers will be recruited to provide lectures in a variety of AI topics that our local AI SME’s may not be knowledgeable about.  
 
Classes are planned to be structured and based off of the textbook Artificial Intelligence: A modern Approach by Stuart Russell and Peter Norvig. This textbook is nicknamed the “Bible for AI” by Doug Lange (NIWC PAC SSTM AI) and has been adopted by over 1500 universities/colleges for their AI curriculum.  

Novice-level topics planned to be covered include: 
Supervised/Unsupervised Learning 
Deep learning 
Computer vision 
Natural Language Processing (NLP) 
Reinforcement Learning, Q-Learning 
Multi-Agent Reinforcement Learning (MARL) 
Automated Planning/PDDL (Planning Domain Definition Language) 
Adversarial Planning, MCTS (Monte Carlo Tree Search), AlphaZero 
Probabilistic Reasoning 
Robotics 
Explainable AI 
AI Dangers/Ethics 

Whenever possible, following each of the lecture modules will be a Google Colab laboratory exercise for students to work on directly with help from the instructor and NAUTICAL team.  

Learn AI with NAUTICAL

We propose a novel method to estimate the pose of a camera relative to a target of interest using only two keypoint features. This addresses a limitation of classical computer vision methods and many state-of-the-art techniques, which require at least three keypoint features to return a unique solution. Our method can be implemented with a monocular camera using information from camera imagery, intrinsics, and extrinsics like its rotation and altitude. This set of known parameters enables us to derive a unique solution for the camera pose using a trigonometry-based approach. In this work, we use a U.S. Naval ship as our target of interest in a simulated dataset. To show generalizability of our method, we test it against an open source 6D pose dataset containing several household items. We compare the performance of our method against various Perspective-n-Point (PnP) methods on these datasets to evaluate for robustness and accuracy. By matching performance of existing PnP methods on the 6D pose dataset and outperforming them on the maritime synthetic dataset, we have demonstrated that our method is effective, robust, and versatile for use in real world applications.

P2P-Trig: An Efficient 2-Point Trigonometric Solution for Camera Pose Using Known Camera Parameters

Quantum-Resistant Endpoint Security for U.S. Navy Document Integrity
HP is pioneering quantum-resistant endpoint security to protect sensitive U.S. Navy documents from emerging quantum computing threats. Their technology embeds NIST-standard post-quantum cryptographic algorithms into proprietary hardware, including the 5th generation Endpoint Security Controller (ESC) for PCs and custom ASIC chips for printers. This hardware-level protection secures BIOS integrity and device boot processes, preventing firmware attacks from even quantum-enabled adversaries.
By proactively embedding quantum-resistant cryptography, HP mitigates risks like "Harvest Now, Decrypt Later," ensuring Navy documents and communications remain confidential and authentic throughout their lifecycle. The solution also supports secure firmware upgrades, complying with federal mandates such as the Quantum Computing Cybersecurity Preparedness Act and the NIST 2030 PQC migration deadline. This approach future-proofs endpoints with long refresh cycles, affording robust protection for classified and controlled unclassified information.
HP’s poster will showcase their quantum-resistant endpoint security designed to protect U.S. Navy documents from quantum threats. It will highlight their proprietary hardware—5th generation Endpoint Security Controller in PCs and custom ASICs in printers—using NIST-standard post-quantum cryptography to secure BIOS and device boot processes.
In summary, HP's quantum-resistant PCs and printers deliver resilient endpoint security essential for safeguarding Navy mission-critical data and national security against next-generation quantum threats


Premium content

Downloads

Next from NAML 2026

Localization of Aircraft using Several Shipboard Optical Sensors (LASSOS

Similar lecture

Multi-polarization Underwater Object Recognition with Metasurfaces-based Lidar Imagery