AI Researcher | Computer Vision | Machine Learning
tel. +39 *** *** **** | email. [email protected] | www. marcocarletti.it
Scientific Interests. Machine Learning, Computer Vision, Augmented Reality, Computer Graphics, Robotics.
Tech skills. Python, C++, Git, Pytorch, Tensorflow, OpenCV, Linux.
Languages. Italian, English.
👔 Work Experience
- R&D Software Engineer, eVS embedded Vision Systems srl, Verona (IT) - 2019-present
AI Research, Deep Learning, Machine Learning, Computer Vision
Tech: Python, Pytorch, Tensorflow, OpenCV, Git
- Scientific Researcher, Istituto Italiano di Tecnologia, Genova (IT) - 2018-2019
Deep Learning, Computer Vision
Tech: Python, Pytorch, Git
- Software Engineer, eVS embedded Vision Systems srl, Verona (IT) - 2015-2016
Tech: C++, OpenCV, Qt, Git
- Internship, eVS embedded Vision Systems srl, Verona (IT) - 2015
Tech: C++, OpenCV
- Research Assistant, Altair Robotics Laboratory, University of Verona, Verona (IT) - 2012-2015
Tech: C++, OpenGL, OpenCV
- Research Assistant, OpDaTe Physics Laboratory, University of Verona, Verona (IT) - 2012
Tech: C++, Qt
- Vision system for sailboat docking assistance, Embedded Vision Systems Srl
State of the art analysis, hardware selection and software design of a multi-camera system for docking assistance. The project involved 3D reconstruction of the environment and camera localization (SLAM), image color correction and mosaicing. Before deployment, the system was tested in a 3D virtual environment.
Tech: C++, Matlab, OpenCV, Qt, Git, Nvidia Jetson, SLAM, Blender
Role: project leader and lead software developer
- 2D and 3D explainability of deep neural networks, University of Verona
Ph.D. thesis. Describe how visual saliency can be effectively used in the 2D domain (e.g. RGB images) to boost image classification performances. Then, build a 3D active recognition system to consider different views of a target object.
Tech: Python, Pytorch, OpenCV, Git
Role: main author (as PhD student)
- Ultrasound tracking of epatic vessels, University of Verona
Master thesis. Particle filtering approach to perform real-time tracking of epatic vessels in medical ultrasound application. The goal was to compensate the breathing of the patient in order to facilitate autonomous surgery.
Role: main author (as Master student)
- Ph.D. in Computer Science, University of Verona, Verona (IT) - 2020
Machine Learning (Deep Learning), Computer Vision, Active Object Recognition
Title: Saliency-based approaches for multidimensional explainability of deep networks
- Master in Computer Science and Engineering (LM-32), University of Verona, Verona (IT) - 2015
Image Processing and multiresolution analysis, Advanced pattern recognition, Computer vision
Title: Filtraggio a particelle con modelli di riferimento variabili in contesti di tracking biomedicale
- Bachelor in Computer Science (L-31), University of Verona, Verona (IT) - 2012
Signal processing (image and sound), Pattern recognition, Computer graphics
Title: Integrazione di sistemi per l'animazione in tempo reale di modelli antropomorfi
- High School Diploma, Liceo Scientifico G. Galilei (Applied Sciences), Verona (IT) - 2008
Selection of extracurricular courses, schools and activities:
- Visiting PhD student, Istituto Italiano di Tecnologia, Genova (IT) - 2018, 2019
- International Computer Vision Summer School (ICVSS), Università di Catania, Ragusa (IT) - 2017, 2018
- Machine Learning Summer School (MLSS), Universidad Autónoma de Madrid (ES) - 2018
- Saliency-based approaches for multidimensional explainability of deep networks
PhD Thesis, 2020
- Active 3D Classification of Multiple Objects in Cluttered Scenes
ICCV (Workshop), 2019
- 6D Pose Estimation for Industrial Applications
ICIAP (Industrial Session), 2019
- Estimating body fat from depth images: hand-crafted features vs Convolutional Neural Networks
- Analyzing Body Fat from Depth Images
International Conference on 3D Vision (3DV), 2018
- Recognition self-awareness for active object recognition on depth images
- Understanding Deep Architectures by Visual Summaries
- A Robust Particle Filtering Approach with Spatially-Dependent Template Selection for Medical Ultrasound Tracking Applications
- Robust 3D Pose Estimation of a Laparoscopic Instrument with three Landmarks
STAG (Eurographics Italian Chapter), 2015
- Multiscale modelling of surfaces by profilometry based on conoscopic holography
Optical Metrology: Optics for Arts, Architecture, and Archaeology IV, 2013
🚀 Hobbies and Interests
Desire to meet new people, find new challenges and learn new stuff.
Love to play guitar, discuss about cinema, politics, religion, astronomy, physics and science in general.
Not very sporty, but definitely loving hiking and alpinism.