Osian Haines

The research website of Dr Osian Haines, University of Bristol

I am a Research Assistant at the University of Bristol, studying Computer Vision (as part of the Visual Information Laboratory - www.bris.ac.uk/vi-lab)

Vision Guided Locomotion

I am now a research assistant working on a research prohject looking into bioinspired models for vision guided locomotion. Publications so far:

Using Inertial Data to Enhance Image Segmentation - Knowing camera orientation can improve segmentation of outdoor scenes
Osian Haines, Dave Bull and J.F. Burn
International Conference on Computer Vision Theory and Applications, to appear March 2015 (VISAPP '15)
download pdf (1.9 MB)
Data are available here

COGNITO and related work

I was employed for a while on the EU-Funded COGNITO project, which led to the following:

You-Do, I-Learn: Discovering Task Relevant Objects and their Modes of Interaction from Multi-User Egocentric Video
Dima Damen, Teesid Leelasawassuk, Osian Haines, Andrew Calway and Walterio Mayol-Cuevas
British Machine Vision Conference, September 2014 (BMVC '14)
download pdf (via bris.ac.uk)

Multi-user egocentric Online System for Unsupervised Assistance on Object Usage
Dima Damen, Osian Haines, Teesid Leelasawassuk, Andrew Calway and Walterio Mayol-Cuevas
ECCV Workshop on Assistive Computer Vision and Robotics, September 2014 (ACVR '14)
download pdf (via bris.ac.uk)

PhD: Plane Detection from Single Images

This section gives a brief outline of my PhD work, on using single images to detect planes by learning from prior knowledge.

Recognising planes in a single image
Osian Haines and Andrew Calway
IEEE Transactions on Pattern Analysis and Applications (TPAMI) (to appear)
Preprint here on IEEE-Xplore

PhD Thesis: Interpreting the Structure of Single Images by Learning from Examples
Osian Haines, October 2013
My complete, final, corrected PhD thesis is now available
download pdf (18.5 MB)

Visual mapping using learned structural priors
Osian Haines, José Martínez-Carranza and Andrew Calway
International Conference on Robotics and Automation, May 2013 (ICRA '13)
download pdf
Download video (mp4, 9.4MB), or a low quality version (mp4, 3.8MB)
View the conference poster(s)

Detecting planes and estimating their orientation from a single image
Osian Haines and Andrew Calway
British Machine Vision Conference, September 2012 (BMVC '12)
download pdf
Extended one-page abstract
Conference poster

Estimating planar structure in single images by learning from examples
Osian Haines and Andrew Calway
International Conference on Pattern Recognition Applications and Methods, Feb. 2012 (ICPRAM'12)
download pdf

Estimating planar structure in single images by learning from examples - Technical Report
Osian Haines and Andrew Calway
University of Bristol Technical Report CSTR-11-005, February 2012
(Extended version of the ICPRAM paper)
download pdf

See also my list of publications on the Bristol Computer Science page: Osian Haines' publications or my Google Scholar Profile

Code and data

Visual Guided Locomotion

The dataset for the VISAPP paper is available here: Download here (includes readme file)
This comprises six sets of images, from different video sequences, gathered with a head mounted camera. Each image is equipped with its orientation from the Oculus Rift inertial sensor.


I am planning to make as much of my code and data available as possible, so that people can make use of my research and perhaps replicate the results. The code will take some time to make ready; in the mean time some data sets are available, with limited annotation. These are .bmp images of urban scenes accompanied by text files describing the plane/non-plane locations.

Training Set ("University") ~430 images of planar and non-planar scenes Download .zip, 75MB
Test Set ("City" - totally independent of the trainign/cross-validation data set) Download .zip, 24MB
Explanation of data format for the dataset: TruthImagesReadme.txt

Teaching Assistant

I have also worked as a teaching assistant on various courses, including the MSc Programming in C course (website), the Introduction to C++ service course for undergraduate engineering students(website), and more recently Tilo and Andrew's wonderful Image Processing and Computer Vision course (website)

Oscar's reference implementation for the Floodit assignment can be found here: floodit.c . If you are on the C course please have a look at this as it provides useful tips for how to code sensibly and robustly


Personal web page www.osianh.com

LinkedIn profile

The Visual Information Laboratory www.bris.ac.uk/vi-lab