ISC-Perception selected for a Buildings cover
The cover visual brings together the synthetic, CAD and real-workcell sources behind ISC-Perception and the robotic assembly setting the dataset is designed to support.
News
A short record of publications, project milestones and technical work. Longer-form technical writing remains on samueladebayo.com.
The cover visual brings together the synthetic, CAD and real-workcell sources behind ISC-Perception and the robotic assembly setting the dataset is designed to support.
The paper introduces a hybrid vision dataset combining real, photorealistic and synthetic imagery for robotic assembly of Intermeshed Steel Connections.
ConPose presents a jointly trained, single-pass RGB framework for component detection and pose estimation in robotic structural assembly.
Through ISx4 AI Research, I co-authored practitioner-facing work on trace-level observability for production LLM applications and operating patterns for lean AI product teams.
A first-author paper on self-supervised facial representation learning and patch-aware gaze estimation.
QUB-PHEO provides a multi-view visual dataset and benchmark for intention inference in collaborative assembly, covering 70 participants and 36 subtasks.