Structural MonitoringPublished result
Infrastructure displacement monitoring
A computer vision system for structural health monitoring that tracks marker pose and measures displacement against controlled reference motion.
Arman Neyestani, PhD
Computer Vision Engineer & Researcher
Vision systems built to measure the physical world.
I build camera-based systems for structural inspection and displacement measurement, alongside work in visual localization and environmental sensing.
Research Fellow · CeSMA, University of Naples Federico II
Camera geometry
Machine perception
Reference validation
Uncertainty
Selected work
Built, measured, validated.
Published studies, open datasets, and repository outputs are labelled with results, methods and clear limits, starting with my structural measurement work.
Structural MonitoringRepository output
Crack verification
A shared ResNet-101 encoder learns a 128-dimensional embedding for deciding whether two images show the same crack pattern.
Structural MonitoringPublished result
Crack segmentation
Computer vision for crack detection and segmentation in civil infrastructure, using a YOLOv8 workflow and UAV-acquired imagery.
Visual LocalizationDataset + benchmark
SUBVO
An open, calibrated underwater sequence with measured ground truth and a benchmark-oriented monocular visual-odometry pipeline.
Visual LocalizationPublished result
UAV visual localization
Monocular visual and visual-inertial localization studies that propagate feature, camera, attitude, and altitude uncertainty into UAV position estimates.
Core technical stack
Tools across the vision pipeline.
Tools and methods I have applied across published studies and research prototypes.
- Python
- OpenCV
- PyTorch
- TensorFlow / Keras
- YOLO detection & segmentation
- Metric learning
- Camera calibration
- PnP & 6-DoF pose
- Visual odometry & localization
- Uncertainty analysis
- Jetson
- Raspberry Pi
More computer-vision work
Marine, environmental and applied projects.
Applied Vision · 2023
MAR Rover
A four-class segmentation prototype converts orchard imagery into a visible soil corridor and image-space guidance signal for an amphibious rover.
↗
Applied Vision · 2025
Blue crab measurement
Instance masks, aligned depth, and camera calibration are combined to estimate projected crab dimensions in millimeters.
↗
Environmental Sensing · 2024
Wave-height forecasting
A GRU model turns 16-step marine observations into a one-step-ahead significant-wave-height estimate for digital-twin prototypes.
↗Working method
From pixels to physical evidence.
01 Frame the measurement problem.
02 Build the perception and geometry pipeline.
03 Compare against reference data.
04 Report uncertainty and failure modes.
Selected publications
Peer-reviewed context.
Vision-Based Displacement Measurement for Structural Health Monitoring: A Metrology-Oriented Review of Uncertainty Quantification
Buildings 16(13), 2659
Development and Evaluation of a Novel Marker-Based Tracking System for 3D-Scaled Masonry Models Using DeepTag
XXXIII International Scientific Conference Electronics (ET)
SUBVO Dataset: Analyzing Feature Extraction for Underwater Monocular Visual Odometry
IEEE I2MTC
A Low-Cost Vision-Based Monitoring System for Dynamic Testing of a Cantilever Beam
IEEE MetroSustainability
Selected testimonials
Professional Recommendations
Selected testimonials from supervisors and professional collaborators.
Current Research Supervisor
I have worked with Dr. Arman Neyestani in an applied research context involving structural monitoring. His expertise in computer vision, visual measurement, and uncertainty assessment is outstanding. He combines strong algorithmic skills with rigorous metrological thinking, works autonomously to identify effective technical solutions, and consistently connects research outcomes with real-world engineering needs.
Get in touch
Have a vision problem worth solving?
Available for computer vision engineering roles, research collaborations, and focused project work.