Systems registry
researchgithub

Underwater Crack Detection

Applies computer vision to inspection tasks where manual review is costly, risky, or inconsistent.

Inspection Images
Image Prep
Detection Windows
CNN / YOLO
Crack Review

system visual

Protected or source media unavailable. Architecture preview used instead.

System Overview

Underwater concrete crack detection system using CNN/YOLO-style inspection workflows for structural safety.

Capstone/research system.

Implementation Signals

PythonCNNYOLOv3DarknetOpenCV

System flow

A compact view of how inputs move through processing, orchestration, validation, and output.

01

Inspection Images

02

Image Prep

03

Detection Windows

04

CNN / YOLO

05

Crack Review

Engineering decisions

Decision Record

Frame inspection as safety assistance

Problem: Manual underwater inspection is limited by risk and visibility.

Approach: Use CNN/YOLO-style detection to identify possible crack regions for review.

Tradeoff: Field deployment requires false-positive controls and environment robustness.

Outcome: A strong bridge between robotics roots and applied perception systems.

Results / Learnings

README

States high accuracy and precision, but no exact numeric value is provided.

Learning

Inspection AI needs robust review and false-positive handling.

Progressive depth

This page keeps the outcome and architecture visible first. Implementation stack, decisions, constraints, and media are available below so technical depth is opt-in rather than forced.

Adjacent systems