BMVA Symposium: Transfer Learning in Computer Vision (TLCV)

Weds 25th Jan 2017, London [submission date 18th Nov 2016]

Chairs:

Dr Dima Damen (University of Bristol), Dr Gabriela Csurka (XRCE) and Dr Timothy Hospedales (University of Edinburgh)

Videos

All videos from the day are now available online

Programme

[Download Programme]

09:00 - Registration + Coffee
09:15 - Welcome from the Chair(s)
09:30 : Transductive TL for Activity recognition - Teo de Campos, Universidade de Brasilia and University of Surrey
10:00 : TL Methodologies - Massimiliano Pontil, Istituto Italiano di Tecnologia and UCL
10:50 - Coffee Break AM
11:20 : Unifying perspectives on knowledge sharing - Timothy Hospedales, University of Edinburgh
11:50 : Learning Feed-Forward One-Short Learners - Luca Bertinetto, Joao Henriques, Jack Valmadre, Philip Torr and Andrea Vedaldi, University of Oxford
12:10 : Active transfer learning for activity recognition - Tom Diethe, Niall Twomey, Peter Flach, University of Bristol
12:30 - Lunch
13:30 : TL for location-invariant activity recognition: An overview - Dima Damen, University of Bristol
13:40 : TL across digital domains: From Photos to Artwork - Peter Hall, University of Bath
14:30 : Exploiting Virtual Worlds and Domain Adaptation for Driving Scene Understanding - Antonio Lopez, German Ros, Laura Sellart, Joanna Materzynska, David Vazquez, Gabriel Villalonga, Computer Vision Center, Barcelona
14:50 - Coffee Break PM
15:30 : Learning to transfer: transferring latent task structures and its application to person-specific facial action unit detection - Timur Almaev, University of Nottingham
15:50 : Cost-based Feature Transfer for Vehicle Occupant Classification - Toby Perrett, University of Bristol
16:10 : Trace Norm Regularised Deep Multi-Task Learning - Yongxin Yang, Queen Mary University of London
16:20 : Open Discussion on future challenges and opportunities in TL
17:00 - End of meeting

Call for Papers

[Download CFP]

The Computer Vision community is in need of moving beyond dataset or task-specific methods towards those that can efficiently adapt to new tasks or domains in a supervised, semi-supervised or unsupervised manner. We aim in this technical meeting to bring together leading researchers, at various levels in their career, with expertise or strong interest in TL for Computer Vision problems, in order to discuss current challenges and propose future directions including potentially establishing a continuous forum or a workshop series.

We are inviting researchers to present short talks and posters that address the motivation, methodologies, challenges and applications of using TL in Computer Vision. These could include ongoing or recently published works, relevant but not limited to: