Vivian Chiciudean
Aerial Perception
LinkedIn
GitHub
Research in
perception
I am a Ph.D. student and AI researcher at the Technical University of Cluj-Napoca, Romania, conducting research at the intersection of computer vision, deep learning, and autonomous systems. My research focuses on scene perception and understanding for UAVs and autonomous platforms operating in complex real-world environments. I am the first author of UAVid++, an aerial semantic segmentation benchmark with higher-quality labels and an expanded class taxonomy, published in IEEE Transactions on Geoscience and Remote Sensing.
Alongside my research, I serve as a teaching assistant for Computer Programming, Computer Architecture, and Image Processing. I also supervise undergraduate research theses, with a primary focus on software development for AI systems.
In addition, I contribute to the academic community as an active reviewer for leading IEEE journals and conferences in autonomous driving, intelligent transportation systems, and computer vision, including IEEE TIV, IEEE JSTAR, IEEE IV, and IEEE ITSC.
Publications
Personal Publications
Journals
UAVid++: Higher-Quality Labels and Expanded Semantic Taxonomy for Aerial Semantic Segmentation
IEEE Transactions on Geoscience and Remote Sensing, 2026Data Augmentation for Environment Perception with Unmanned Aerial Vehicles
IEEE Transactions on Intelligent Vehicles, 2024Conferences
Boosting Small Object Segmentation and Generalization of DINO-Based Vision Transformers in Aerial Imagery
IEEE ITSC 2026Links will be added after the presentation.
Localization and Change Detection Through Aerial Environment Perception
IEEE ICCP 2024Static Mesh Enrichment with Dynamic Entities for Training Sets Generation
IEEE ICCP 2023Pathfinding in a 3D Grid for UAV Navigation
IEEE ICCP 2022Supervised Student Publications
Co-authored works supervising student research @ TUCN.
Conferences
Semantic Segmentation from Object Detection using Foundation Models
A. M. Schiau, V. Chiciudean, F. Oniga, S. Nedevschi · IEEE ICCP 2025Open-World Video Panoptic Segmentation for Aerial Perception using Foundation Models
N. G. Boncea, V. Chiciudean, F. Oniga, S. Nedevschi · IEEE ICCP 2025Code & datasets
UAVid++
An aerial semantic segmentation benchmark with higher-quality labels and an expanded class taxonomy, together with a training and inference protocol built on frozen DINO-pretrained ViT backbones with lightweight trainable heads.
View project page →
Dataset Augmentation
An approach to reduce the number of manually annotated images required for training in aerial perception.
View on GitHub →
Solar System Game Engine
A C++ game engine from scratch - simulates planetary orbits, spaceships, aliens, visual animations and audio effects. The goal: save the solar system from an incoming meteorite.
Laboratory teaching
Computer Programming
Year 1 · Semester 1 · Laboratory
Computer Architecture
Year 2 · Semester 2 · Laboratory
Image Processing
Year 3 · Semester 1 · Laboratory