1 of 6

2 of 6

CODE VERSE HACKATHON 2025

Atharva Patwardhan

Atharv Muchandi

Vedant Nagmoti

Vishruti Mohinkar

Problem Statement Title - PS1 Marine Fouling

Team Name - Not Like Us

Team Members-

3 of 6

IDEA TITLE

PROBLEM STATEMENT

Ships and underwater structures often get covered with algae, barnacles, and other organisms. This slows them down, increases fuel use, raises costs, and can even harm the environment. The Navy needs a smart system that can look at images, figure out what kind of fouling is there, how much of it has grown, and predict when cleaning or maintenance will be needed.

​

    • AI-powered image analysis – Our system automatically detects type and density of fouling from the hull images.
    • Predictive maintenance forecasting – We predict growth trends and suggest the right time for cleaning saving millions of rupees.
    • AR-enabled 3D ship model – Our immersive 3-D Model shows fouling hotspots on an interactive 3D hull for easy visualization and density mapping
    • Risk & cost impact insights – Our intelligent system translates fouling into drag %, extra fuel use, CO₂ emissions, and cost for profits prediction.
    • Scalable & software-first – Works without heavy hardware, uses synthetic data, and can scale across fleets.

INNOVATION IN OUR SOLUTION

PROPOSED SOLUTION

1

2

3

4

5

6

7

Image Capture with ROV + UV Camera

An underwater ROV equipped with RGB + UV camera will scan the ship hull and captures clear, geo-tagged images.

Preprocessing & Enhancement

Images are corrected by our software for underwater distortions using dehazing, denoising, and color correction for better visibility.

​

Fouling Segmentation & Density Detection

Deep learning models (CNN/U-Net) identify fouling regions and calculate density percentage across the hull surface.

Species & Growth Stage Classification

The AI classifier labels fouling types (algae, barnacle, slime) and stages (early, medium, severe).

Fouling Index & Risk Scoring

A combined index evaluates drag, fuel penalty, and operational risks to prioritize critical areas.

Predictive Maintenance Engine

Time-series forecasting predicts future fouling growth and suggests optimal cleaning schedules to minimize costs.

3-D Hull Mapping & AR Visualization

Fouling hotspots are overlaid on a 3D digital twin of the hull, viewable in AR for immersive inspection.

4 of 6

TECHNICAL APPROACH

TECH STACK

Frontend

Backend

Computer Vision & AR

OpenCV

Unity

Density Detection

Forecasting

Machine Learning

IoT

Web+DB

DETAILED WORKFLOW

ROV With UV -light and waterproof camera (GoPro)

Python

Express.js

CNN classifier

Tensor Flow

LSTM

+

+

U-Net

5 of 6

FEASIBILITY AND VIABILITY

POTENTIAL CHALLENGES

    • Unclear Underwater Visibility → We use ROV-mounted UV/LED lighting and AI-based image enhancement (dehazing, color correction) to capture clear and reliable images.
    • Limited Training Data → We use synthetic fouling images, data augmentation, and transfer learning to improve AI accuracy across species and growth stages.
    • Real-Time Integration & Scalability → We use lightweight, optimized AI models on edge devices and a modular design to handle detection, prediction, and AR visualization efficiently.

​

Unclear Underwater Visibility

Dark waters, biofilm, and poor lighting reduces image clarity, making fouling harder to detect accurately.

​

Limited Training Data for Biofouling

Few publicly available datasets of algae and barnacle species → risk of lower AI accuracy without synthetic/augmented data.

​

Real-Time Integration & Scalability

Running AI detection, predictions, and 3D AR visualization together in real-time on low-power systems is resource-intensive.

​

STRATEGIES TO OVERCOME CHALLENGES

​

    • Target users: Indian Navy, Coast Guard, commercial shipping companies, offshore platforms.
    • Problem scale: Biofouling increases fuel costs by 10–30% and causes maintenance delays globally. Thats why our system is a game changer
    • Software-first approach: Low-cost, scalable, and easy to deploy across multiple ships and fleets.

​

    • Low hardware cost: Uses affordable ROVs and cameras; no need for full dry-docking.
    • Operational savings: Reduces fuel consumption, maintenance time, and labor costs.
    • Scalable revenue model: We will provide SaaS (Software as a Service ) subscription for fleets, optional AR dashboard upgrades, predictive maintenance analytics as a service.

MARKET FEASIBILITY

FINANCIAL VIABILITY

6 of 6

IMPACT AND BENEFITS

PROBLEMS IN TRADITIONAL METHODS

    • Manual hull inspection by divers

IMPACT & BENEFIT

    • Our ROV (Remote Operated Vehicle) + AI system makes inspection faster, safer, and consistent without diving.
    • Visual inspection by eye is not that accurate to predict algaes.
    • Our intelligent AI detects species, density, and stage accurately.
    • Randomized maintenance approach
    • No digital record or any paperwork
    • No 3D visualization tools or mapping systems
    • High manpower & operational cost
    • This leads to high fuel costs and unexpected downtime so,our predictive maintenance optimizes cleaning schedules and saves fuel.
    • Hard to track fouling growth; Our system creates digital hull records and trend analysis
    • Operators cannot see hotspots clearly therefore our AR 3D hull model highlights fouling locations for targeted cleaning.
    • Maintenance requires many divers and repeated inspections; on the other hand our proposed Software-driven solution reduces manpower, cost, and effort.