Alastair Barber, Research Engineer
In July, I was lucky enough to be accepted to attend the International Computer Vision Summer School (ICVSS) to be held in Calabria, Italy. Of course, the combination of words ‘Summer’ and ‘School’ have been known to induce panic in many people – however, what followed were five days of excellent and insightful talks, demonstrations and conversations over drinks with highly regarded members of the computer vision community, from across industry and academia. Aimed squarely at first year PhD-level students, it was also a fantastic opportunity to get to meet those who were in a similar position to myself from around the world and who would be the ones leading the field forward in the future.
The school is held annually and directed by Sebastiano Battiato and Giovanni Maria Farinella of the University of Catania (Sicily, IT) and Roberto Cipolla from the University of Cambridge (UK). It was started with the intention of providing a friendly and supportive environment for early career researchers to get up to speed with both the theory and practice of the cutting edge in computer vision research. Normally held in Sicily, unfortunately, for 2013 the school was forced to move at the last minute from its usual location in Sicily, to Calabria, some 300 miles away! Whilst this did mean a rather extended coach journey leaving at a time I’d only want to see once in the day, the location was just as stunning:
The view from the resort. Not bad!
The loose theme for 2013’s session was ‘Computer Vision and Machine Learning’. To this end, there was a focus on algorithms for image segmentation and also in new machine learning techniques – in particular Deep Learning and Decision Forests. Of course, Computer Vision is a wide and diverse domain, and as such everyone in attendance would have had differing levels of experience and research interests. Recognising this, all of the lectures were structured in such a way as to give a good introduction to an overview of the problem that the speaker was trying to address, before really going down into quite low level details of the subject. Through this approach, and the excellent quality of speakers, I found myself developing a deep understanding of many subjects and methodologies that were quite far from my particular research interest. Overall, the quality of the talks was excellent, and it was also great to hear from speakers both in academia and industry. This mix gave a good idea of how ideas from academia could be adopted by industry and to create successful products (one such example being the Microsoft Kinect) and also on the challenges that are faced in a practical real-world environment.
Another great aspect of the school was the encouragement given to participants to present their work and research ideas. The poster session that was part of the event was not only a great chance to present and discuss your own work with others but also a great way to find out about what others were researching and also different approaches to tackle similar problems. One poster in particular [‘Towards A Visual Gyroscope’ – Hartmann W., Havlena M., Schindler K. – ETH Zurich] provided a very relevant insight into a different method of accomplishing something I’ve been working on extensively. Naturally, the opportunity to meet and discuss your work with others wasn’t just limited to formal sessions. One of the stated aims of the organisers, and their decision to hold the school in a remote location within a self-contained resort, was that we were to get to know one another in our own time, make friends and contacts and generally interact with one another as opposed to sticking in already-formed groups. To this end, parties on the beach, late night pool sessions and plentiful sampling of various local beverages were commonplace throughout the week!
Party on the beach at night
All in all, the school was an amazing experience, and I am extremely grateful for the organisers for putting on such an event. As I mentioned to one of the organisers before leaving, I didn’t sleep much throughout the week, and that by no means was a bad thing! If anyone is reading this thinking about going – then you should definitely apply! The opportunity to meet people from all over the world at various stages of their career and share experiences and knowledge is wonderful – I’m still in touch with several of the people I met and will no-doubt be seeing them again at some point at conferences.
I’ve just started my second year of the EngD program, working with Double Negative Visual Effects (www.dneg.com). My main research area is camera tracking with a vision to provide accurate real-time information on camera movement during filming for live-preview of visual effects. To this end I’ve been looking at various hardware technologies (such as gyroscopes and accelerometers) and computer vision techniques.
Double Negative is a multi-award winning visual effects studio that has grown to become Europe’s largest VFX provider for feature film. Recent work has included, amongst others, The Dark Knight Rises, Bourne Legacy, Skyfall, Les Miserables, Fast & Furious 6 and Man of Steel.