Postdoc in texture synthesis at UCL

This job posting is expired. For more recent job openings in Computer Vision please visit our jobs section.

A postdoc position is available in the vision group at University College London. Funding is available for one year, in the first instance, to work on Structured Texture Synthesis generally, and specifically on synthesis of realistic-looking handwriting.

We seek an excellent candidate with a track record of expertise in some combination of computer vision, machine learning, and computer graphics. The post is for someone smart and really driven, who will get involved in our diverse vision and graphics projects, but with primary focus on and responsibility for this EPSRC-funded project. Everyone from all nations that matches the Person Specification criteria is encouraged to apply, and the post must be taken up BEFORE March 2013.

Dr Gabriel Brostow's group, based at UCL in the center of vibrant London, performs research in computer vision and data-driven graphics. The group's specialty is Smart Capture for Vision-System Authoring, where systems must model their uncertainty, and should ideally improve (themselves) with experience. This project has a significant Active Learning component, because the uncertainty here includes modeling of each human's handwriting style, and measuring the informativeness of different text-snippets for which the human could be prompted.

This is a precursor-project, leading both to further research in structured texture synthesis (where the user wants to control easy/big parameters, but wants the details of ink, 3D shape, video etc. to be synthesized "for free"), and to further projects in forensic computer vision. This post-holder will need to interact with several forensic handwriting experts, and will co-organize a 1-day workshop for vision and forensics experts.