Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 39 for “"Scene Representations"”.
-
Predictive scene representations for embodied visual search
… the world using sensory inputs, build rich representations of their surrounding environment and take actions to perform their tasks. Augmented reality (AR) assistants must perceive activities performed by humans to provide assistance. My dissertation tackles embodied visual search, where the …
-
Scene Representations for Generalizable Novel View Synthesis
… synthesis involves generating novel views of a scene when seen from different viewpoints. It offers numerous applications in computer vision domains such as telepresence, virtual reality, re-cinematography, etc. Recent literature work in the field successfully achieved remarkable photo-realistic …
-
Improved Stixels Towards Efficient Traffic-Scene Representations
Stixels are medium-level data representations used for the development of computer vision modules for self-driving cars. A stixel is a column of stacked space cubes ranging from the road surface to the visual end of an obstacle. A stixel represents object height at a distance. It supports object …
-
Real-time 3D scene representations for robotic systems
Real-time 3D scene representation is a fundamental capability for robotic systems operating in dynamic and resource-constrained environments. In this thesis, “real-time” refers to perception and reconstruction pipelines operating under update latencies ranging from milliseconds to seconds, …
-
From 3D Mapping to Scene Representations for Embodied AI
… various challenges, including the development of scene datasets and simulators used to train AI agents in diverse tasks, necessitating a comprehensive set of skills. Generally speaking, Embodied AI research projects assume a similar setup where an agent equipped with a set of sensors (usually an …
-
Differentiable 3D Scene Representations with Point-based Neural Methods
… thesis explores particle-based approaches for scene representation. We introduce moving patches by fitting geometry pixels ("gexels") of moving geometry images onto the scene geometry. We propose to represent a signed distance field as a weighted sum of local functions attached to moving …
-
Active Perception for 3D Scene Representations in Robotics from an Information Theoretic Perspective
… perception systems utilizing radiance field representations, particularly 3D Gaussian Splatting, from an information-theoretic perspective. We first address the challenge of quantifying observed information in 3D scene representations. By leveraging the Fisher Information Matrix (FIM), we …
-
Data-driven Spatially and Temporally Consistent Scene Representations for Robotic Minimally Invasive Procedures
… and structured understanding of the surgical scene. This thesis addresses this limitation by investigating how visual information in robotic MIP can be organised into scene representations that remain coherent across space and time. The central premise is that surgical understanding depends …
-
Scene Representations and Data-driven Priors for 3D Reconstruction and Rendering in the Wild
… Autonomous vehicles, for instance, must build 3D representations of their surroundings to navigate and interact safely with other traffic participants. Beyond perception, developing and validating such systems demands realistic closed-loop simulation, since safety-critical scenarios far exceed …
-
The Perceptual Specificity of Representations of Complex Scenes for Rotation, Translation, and Reflection
The issue of how complex visual scenes are represented has received relatively little attention. Perceptual specificity refers to the degree to which a physical change to a stimulus, upon re-presentation, alters performance in any type of memory test. I investigated the perceptual specificity of …
-
Learning to represent scenes with geometry and semantics
<p>Scene representation is the process of converting sensory observations of an environment into compact descriptions. Such intelligent behavior is a cornerstone of artificial intelligence. Scientists have long sought to reproduce the extraordinary ability of humans for understanding the physical …
-
3D reconstruction and scene understanding with 3D Gaussian Splatting representation
… lies the need for efficient, expressive 3D representations capable of capturing geometry, appearance, and semantics. While recent neural scene representations have notably improved image-based reconstruction, they are often hindered by high computational costs, limited robustness, and poor …
-
Language-Driven Robotics: Dynamic Closed-Loop Control for Complex Manipulation
… changing environments. By leveraging advanced 3D scene representations, vision-language models, and innovative methodologies, the research addresses the challenges of grasping and manipulating novel objects in real-time. The proposed approach combines Gaussian Splatting, a fast and efficient 3D …
-
Label-efficient learning paradigms for scalable robot perception
… of their surroundings. These structured scene representations support key downstream tasks such as path planning, motion forecasting, and collision avoidance. However, learning such representations typically relies on dense and expensive forms of supervision such as 3D occupancy grids, …
-
Understanding word meaning through scenes
… in the context. This dissertation proposes the Scene Abstraction Framework (SAF), a computational framework for representing word meaning in relation to the situations in which words occur. SAF draws on Fillmore's Frame Semantics as its theoretical foundation and uses an instruction-tuned …
-
Learning for Robot-centric Autonomy
… visual experiences into sparse, topological scene representations for long-horizon navigation; (2) At the semi-local level, we develop perception systems that reason about the traversability of the terrains around the robot to achieve robust off-road navigation; (3) Locally, we learn …
-
Parametrization, auralization, and authoring of room acoustics for virtual reality applications
… the sound and the visual parts of the virtual scene are in coherence with each other, which should improve the user immersion in the virtual environment. The second aim was to verify the feasibility of the created sound environment parameter set in practice. A virtual acoustic modeling system …
-
Annotation-free Learning for Sensor Fusion in ADAS
… methods for feature matching between scene representations to improve model performance.
Page 1 of 2