Foto del docente

Luigi Di Stefano

Professore ordinario

Dipartimento di Informatica - Scienza e Ingegneria

Settore scientifico disciplinare: IINF-05/A Sistemi di elaborazione delle informazioni

Didattica

Ultime tesi seguite dal docente

Tesi di Laurea

  • Analisi di prodotti tramite visione artificiale per processi automatizzati
  • Analisi sperimentale dell’accuratezza di posizionamento di una telecamera stereo dotata di IMU
  • Creazione di un dataset per lo sviluppo di una rete neurale che esegua la segmentazione del teschio umano in cranio e mandibola
  • Realizzazione di una rete neurale convoluzionale per la classificazione di immagini
  • Sensori di immagini: la tecnologia HDR (High Dynamic Range)
  • Sensori LiDAR: uno studio sul funzionamento della tecnologia e sull’utilizzo in campo Automotive
  • Vision Language Models as Image Classifiers: an Experimental Study

Tesi di Laurea Magistrale

  • A neural network approach to automatic color correction for produce identification
  • A study of 3D face reconstruction methods: from geometry to deformable mesh models"
  • A Test Time Adaptation Protocol to improve Industrial Anomaly Detection and Segmentation
  • Advancements in Industrial Anomaly Detection: From SOTA Models to Online Classification and Segmentation
  • Advancing Car Insurance Fraud Detection: An Automated Deep-Learning Approach
  • An Innovative Approach for Classifying and Distinguishing Bubbles in Pharmaceutical Liquids
  • Analisi di immagini iperspettrali di differenti tipologie di polimeri tramite tecniche di Machine Learning
  • Analysis of the classification performance of Produce Image recognition exploiting automatic color correction network and PLU produce hierarchy for training a multi-stage classifier
  • Applicazione di tecniche AI per la segmentazione di immagini industriali con supervisione limitata
  • Automated Blender environment recreation and vanishing point detection for image stitching optimization
  • Automated Pain Assessment from Facial Expressions of Elderly People using Machine Learning
  • CAD-based Image Generation and Template Matching for tyre manufacturing process
  • Comparison between traditional vision and AI-based inspection systems for crimping quality checking in aseptic pharmaceutical manufacturing
  • CVAT meets transformers: accelerating semantic segmentation labeling in industrial applications
  • Deep Learning approaches for multimodal analysis in car collision forensics
  • Deep Learning-based Classification and Detection of Regulatory Elements for Autonomous Driving Maps in BEV View
  • Deformazioni superficiali automatiche di un oggetto modellato tramite Neural Radiance Fields
  • Deployment of event cameras for parcel profilometry
  • Form extractor: information extraction from digital documents
  • Image-Guided Wheat/Corn Kernel Detection with OWL-ViT and YOLO-WORLD
  • Implementation of Optical Character Verification using Deep Learning for Detection and Verification of Tyre Markings
  • Improving reliability and effectiveness of decoding libraries
  • Individuazione di difetti in manufatti industriali a partire da modelli sintetici
  • Integrating AI models into Unreal Engine
  • Integrating Large Language Models into Extended Reality for Enhanced Educational Experiences
  • Integration and optimization of a tablets uniformity detection system with AI techniques
  • Mask R-CNN Based System for Accurate Detection and Analysis of Chip Damage on Car Paint
  • Methods and Tools for 3D Rendering with Gaussian Splatting in car accident reconstructions
  • Modelli neurali per industrial anomaly detection
  • Multi-view 3d reconstruction using NeRF-based approaches
  • Night2Day - AI enhancement of night photography
  • Optimization of Stereo Reconstruction algorithms and AI models for real-time on-edge compute in complex Minimally Invasive Surgery
  • Patchcore Transformed: A Vision Transformer Approach To Anomaly Detection
  • Piattaforma Integrata per la Ricostruzione 3D di Incidenti Stradali: Dall'Acquisizione alla Manipolazione e Visualizzazione dei Modelli
  • Produce recognition: a deep metric learning approach to an open set learning problem
  • Real time parcel tracking and image stitching for industrial applications
  • Relighting Neural Radiance Fields Leveraging Shadow Mapping
  • Ricerca mediante immagini e testo di asset 3D rappresentati come Neural Radiance Fields
  • Rilevamento centro corsia mediante metodi di apprendimento automatico
  • Rilevamento di intrusioni in tempo reale mediante video di sorveglianza a bassa risoluzione e riconoscimento facciale
  • Rilevamento e classificazione delle specie arboree in ambito urbano utilizzando immagini satellitari ad alta risoluzione
  • Robotic Bin Picking and Quality Control: Computer Vision Approaches for Industrial Automation
  • Soil Organic Carbon mapping using hyperspectral satellite data
  • Sperimentazione di architetture Transformer in problemi di Computer Vision
  • Sviluppo opto-meccanico e del suo algoritmo per un sensore interferometrico
  • Synthetic Data Augmentation For Tabular Data Via Deep Learning Methods
  • Towards image-guided additive manufacturing
  • Uncertainty-aware learning-based optimal motion planning for industrial robots
  • Use of artificial vision libraries in industrial applications for quality control in the production cycle
  • Visual Servoing Meets LQR: Streamlining Automated Cliché Mounting with Computer Vision and Precise Three-Axis Control

Tesi di Dottorato

  • Learning with limited data