Ben Canini
About
My research project
Sketch 3D: Sketch-Based Methods for 3D Recognition and LocalisationMy project is concerned with interaction between humans and 3D environments through the medium of sketch. Many robotics, AR, and VR applications currently operate using photorealistic images or text input, while abstract human sketches remain relatively underexplored - especially in 3D.
Sketch is a powerful tool for creating a visual representation of something that may otherwise be difficult to describe, such as a specific viewpoint, shape, or instance of an object. My work focuses on two key areas: representation - what is the best way to represent a 3D environment to facilitate interaction through sketch? And localisation - from where in a 3D environment was a sketch drawn? Or, given a sketch of an object, where is that object located within the environment?
Together, these areas aim to make sketch a practical interface for interacting with and reasoning about 3D environments, exploring how abstract, human-centred inputs can complement conventional image and text-based approaches.
My project is concerned with interaction between humans and 3D environments through the medium of sketch. Many robotics, AR, and VR applications currently operate using photorealistic images or text input, while abstract human sketches remain relatively underexplored - especially in 3D.
Sketch is a powerful tool for creating a visual representation of something that may otherwise be difficult to describe, such as a specific viewpoint, shape, or instance of an object. My work focuses on two key areas: representation - what is the best way to represent a 3D environment to facilitate interaction through sketch? And localisation - from where in a 3D environment was a sketch drawn? Or, given a sketch of an object, where is that object located within the environment?
Together, these areas aim to make sketch a practical interface for interacting with and reasoning about 3D environments, exploring how abstract, human-centred inputs can complement conventional image and text-based approaches.
Publications
Robust 3D edge reconstruction from in-the-wild data remains a key challenge in robotics and SLAM, where scenes are often noisy, unstructured, and captured under imprecise conditions. Recent methods based on neural implicit representations offer compact reconstructions but suffer from high computational costs and poor localization of fine edge detail due to reliance on NeRF-style volumetric rendering. While Gaussian Splatting offers a fast and accurate alternative, its use in edge extraction has been limited to clean or synthetic datasets with precise initialization.We present EdgeSplats, a method for 3D edge reconstruction that operates effectively on real-world data without requiring clean geometric priors. Our approach leverages the 3DGS-MCMC pipeline to train edge-aligned Gaussians directly from posed images. A graph-based clustering stage then extracts edge-consistent splats by exploiting spatial and directional coherence.We demonstrate that EdgeSplats improves both reconstruction quality and training speed over existing methods, demonstrating strong performance on noisy, real-world inputs.