Rosario Scalia

Comptuer Vision Engineer

Xenia Progetti•  October 2021 - Present

[ DESCRIPTION ] 1) Software development that allows the automatic detection of accidents in the context of road construction sites. The developed system exploits the video streams of cameras to identify the following types of accidents: - Worker on the ground - Worker without PPE (helmet, vest) - Vehicle-man distance too low - Vehicle-equipment distance too low - Worker outside the perimeter. The solution was developed and optimized for NVIDIA Jetson Xavier and exposes a set of REST APIs that allows on-demand checking about accident events as well as allowing the view of video stream ,with accidents reporting, on a Web Browser. In addition, the solution automatically forwards alarms to a remote backend together with the automatic send of an SMS, containing the description of the accidents, to a predefined list of phone numbers. 2) Development of Augmented Reality application for museum contexts on Nreal Light headset. [ ACQUIRED SKILLS ] - Deep Learning models for pose estimation - Deep Learning models for object detection suitable for embedded devices - Deep learning model optimization for NVIDIA Jetson Xavier throught TensorRT API - Python backend development with fastAPI - Perimeter abandonment detection through homographs - Vehicle-Object distance detection via depth camera (RealSense and Zed Cam) - Design Patterns that allow you to package solutions in a modular pipeline - Interprocess communication via ZMQ API - Video tracking algorithms - Man on the ground detection through algorithms that process the poses detected by the pose estimator - Development AR apps on Unity - Unit Test Libraries for C# - Assembly concept and modular compilation on C# - Development on Nreal Light headset - WebSocket communication protocol

Computer Vision Engineer

DWORD•  March 2020 - October 2021

[ DESCRIPTION ] 1) Software Development for Automatic Detection of Individual Protection Devices through videos processed by Machine Learning Algorithms. The application context was a refinery managed by a customer of the company. 2) Software Development for Automatic Detection of Interpersonal Distances from video acquired by cameras. The application context was the fight against the COVID-19 virus, The software solution was industrialized for some Company's customers. [ ACQUIRED SKILLS ] 1) Machine Learning models for the Object Detection 2) Computer Vision algorithms for calculating the distances between People in a Scene

Augmented Reality Developer

IP Lab - Unict•  October 2019 - January 2020

[ DESCRIPTION ] Development of an Indoor Localization Systems on Microsoft Hololens. Software is based on a Neural Model executed locally on the device. [ ACQUIRED SKILLS ] 1) Development on Microsoft Hololens (Unity and UWP) 2) Development with the C # Programming Language 3) Exporting Neural Models in a cross-framework format through ONNX

Intership on Computer Vision

Xenia Progetti•  May 2019 - June 2019

[ DESCRIPTION ] Software development that allows the localization of a User inside a building. To do this, the analysis of the images captured by a viewer worn by the User was used. (Microsoft Hololens). The analysis algorithms used are based on Deep Learning. [ ACQUIRED SKILLS ] - Building Neural Models with the PyTorch library - Software development in Python - Viterbi algorithm - Graphic interface development with Python - Deep Learning Training Results Analysis

University of Catania

Computer Science, MS•  October 2019 - April 2022

Master's Degree in Data Science & Computer Vision fields