Adrian Clark obtained a BSc in Physics (first class, and winner of the Stroud Lab Prize) at the University of Newcastle upon Tyne in 1979 and a PhD in Image Processing from Queen Elizabeth College (now merged with Kings College; see the Wikipedia for a brief history of QEC), University of London in 1983. After postdoctorate research into parallel image processing on the mini-DAP at Kings College, he worked for British Aerospace (now BAe Systems) during 1985-88, researching in the general area of object recognition. ;-) He joined the Department of Electronic Systems Engineering at the University of Essex in 1988, and that eventually merged with the Department of Computer Science to form the current School of Computer Science and Electronic Engineering. He is now a Reader.
During 1991-93, he acted as Convenor of BSI IST/31/-/6, a panel of experts involved in the development of the International Standard for Image Processing and Interchange (IS 12087), and was editor of the first part of that Standard. He is a former chair of Technical Committee 5 (Benchmarking and Software) of the International Association for Pattern Recognition and a fervent evangelist for the quantitative evaluation and comparison of algorithms in computer vision and image processing, being involved in the organization of all significant events in the area for the first decade of attempts to quantify the performances of vision algorithms, and is active in the use of such measures in designing vision operators and systems.
This interest in performance characterization was one of the major purposes for setting up PEIPA, the Pilot European Image Processing Archive, which provides software, imagery, and support material to the computer vision and image processing communities. He also moderated The Pixel, a bulletin board and email list concerned with computer vision and image processing, for about a decade. In addition to the BMVA, he is a member of the EPSRC College, of the IEEE and the Institute of Physics, where he served on the committee of the Signal Processing, Electronics and Communication Systems group.
He is currently chairman of the British Machine Vision Association, which represents the interests of computer vision and pattern recognition research in the UK, as well as organising a couple of conferences, a summer school and innumerable technical meetings each year. The BMVA is also one of the first fully open-access publishers, having always made its conference proceedings available to the community free of charge; and the Annals of the BMVA is also fully open access.
He worked as Essex's main Business Fellow with the London Technology
Network for a number of years. His interest in transferring research
advances into UK industry remains, through collaborative work under Knowledge
Transfer Partnerships (KTPs), and he is currently involved in two of them.
- School of Computer Science and Electronic Engineering
University of Essex
- +44 1206 872432
- +44 1206 872900
- alien /at/ essex.ac.uk
- Web page
If you're on the Essex campus, my office is in the big white building on Square 1, the lowest of the five main squares. Walk into the building from square 1 and turn left in the foyer, then follow the corridor round to the right, which means you'll be heading away from square 1. Go through the door which leads to the staff offices, and my office is the second-last one on the right. If I'm not in, I usually leave a note above the name panel to the right of my door.
My research explores computer vision and augmented reality — and, increasingly, the combination of the two. Augmented reality involves taking 3D geometry and surface descriptions and producing 2D images of them, superimposing them on the real world; vision often involves the inverse operation, taking 2D images and trying to figure out the 3D geometry that appears in them. The second problem is undoubtedly the harder of the two and circumventing the ill-posed nature of the problem is one of the fascinations of working in the area.
Specific things my colleagues and I are doing in the Vision and Synthetic Environments Laboratory ("VASE Lab"), include:
- Research into the design and construction of vision systems. Work with Olly Oechsle (who won the BMVA's Sullivan thesis prize in 2010) saw the development of a vision system that is able to adapt to totally new problems purely through training. That system was advanced significantly by Panitnat Yimyam in her doctoral research. You're welcome to download and try out Jasmine, Olly's and Panitnat's software which implements these ideas. The underlying machine learning algorithm is genetic programming and Jasmine has been able to solve problems as disparate as classifying skin lesions as benign or cancerous, distinguishing damaged and undamaged fruit and reading car number plates purely by being trained with different data. Some skin lesions are shown below: the left column shows lesions identified as moles, the right column as melanoma. Also shown below is the series of learnt processing stages in reading number plates.
- Hand in hand with this is research into quantitative ways of measuring the performances of vision systems — my interest in this dates back to my time in industry. As the vision research community is now happy with the idea of measuring the performance of algorithms in isolation, we are now taking the next step, comparing the performances of algorithms in a statistically meaningful way. We have also made steps towards integrating the measurement of performance into the design of individual operators, not just complete systems. This is exciting stuff, at least for my students and me!
- The problem with conventional virtual reality (VR) is that it is totally unrealistic: the user either wears shutter glasses and views a monitor (or, if one is exceptionally lucky, stands in a CAVE), or wears an expensive head-mounted display that ties him or her to a fairly beefy computer. A major theme of our recent work is untethered AR; this ties in (ho, ho) rather well with the vision-based research mentioned above as vision is also, of course, inherently untethered. To achieve untethered AR, we need a computer that is small enough to be carried or, better, worn; and so we have developed a theme of research into wearable computing. In particular, we are using a wearable computer equipped with suitable sensors as a wearable augmented reality tour guide for local archaeological sites. (Colchester was the first capital of Roman Britain, and there is a fair number of important archaeological sites dating from that period.) Our current work in this area is developing a system that uses visual SLAM to determine the wearer's position and orientation relative to real-world objects.
- Devising a framework for constructing user interfaces and applications for wearable computers that is able to accept input and deliver output via several media, and which is able to adapt to the user's environment. We believe that this framework, Sulawesi, is the first general-purpose multi-model, contact-aware user interface framework. Sulawesi supports textual or spoken inputs and is able to switch between textual and spoken output on the fly. It also interfaces to position-sensing devices and body-worn movement sensors, and hence is able to provide location-based output (imagine being told Remember to buy some milk on the way home by your wearable as you leave work) and to switch between output modalities depending on the context: for example, Sulawesi normally switches from visual to spoken output when it realizes the user is moving as it's pretty annoying to have a visual interface doing things in front of your eyes when walking or driving! More recently, we have monitored the habits of computer use that people develop, often sub-consciously, and have the machine adapt to them — context-aware computing. The aim here is to make the computer much more proactive, obtaining and organizing the information that user needs, say, in anticipation of it being required.
Do take a look at the VASE Lab projects page to find out more and see what else we're doing.
Over the last year or so, I have converted a defunct research laboratory into a place where we are establishing a new thread of research into Virtual Reality and human-computer interaction. The lab, the Networked Interaction Laboratory or NIL, features an ultra-high-resolution stereoscopic 4,200 × 2,400-pixel display that fills a whole wall — the size of a pixel is 1 mm2! We have had some fun in producing 3D reconstructions of places, the most spectacular of which is Paternoster Row (in London) as it would have appeared at the beginning of the 20th century. Non-traditional forms of interaction include voice, gesture (courtesy of a Microsoft Kinect) and bicycle — yes, you can cycle down Paternoster Row!
Externally-funded research has encompassed areas such as stochastic techniques for spectral identification in remotely-sensed spectra and hyperspectral imagery; model-based image coding; pilot landing aids for civil aircraft; face recognition; and shared virtual worlds using VRML and IP multicasting. All EPSRC-funded projects have received good ratings.
I enjoy working with industry. I am currently engaged in two Knowledge Transfer Partnerships (KTP) with local companies, and am in the process of setting up another.
The Essex system of study ("sabbatical") leave means that I have taught a wide range of subjects over the years, at both undergraduate and postgraduate levels. Subjects I've lectured include (in no particular order): computer vision; digital signal processing; real-time computing; high-performance computing; C, C++, Java, Python, Tcl/Tk; software engineering; parallel and distributed computing; human-computer interaction; operating systems; computer networking; computer security; and communication skills.
My current teaching commitments are:
- CE316: Computer Vision
- CE816: High-Performance Computing
- CE866: Python Programming
- CE866: Computer Vision
I have been active in course and curriculum development right from my early days as an academic, when I had scripting languages form part of our programming and software engineering offerings; I suspect we were one of the first institutions to do this. I developed my School's popular MSc degrees in Computer and Information Networks and in Computer Security and conceived much of their content. I had major involvement in what is now our undergraduate degree in the area of Computer Games, and introduced our undergraduate degree in Computer and Network Security. In the early days of CSEE, I made significant changes to the way in which all the School's MSc degrees operate, as well as introducing some new degrees and rationalising others.
I lecture on the annual EPSRC Summer School on Computer Vision, which I dreamt up in the early 1990s with Bernard Buxton and ran for a number of years. I have also organised a student papers conference for the BMVA, in which papers are written, presented and reviewed by research students.
I currently have no external examining commitments but in the past have been external examiner for undergraduate degrees at De Montfort and Reading; and for postgraduate degrees at Surrey and UEA. I have been involved in assessing and validating degrees at Hertfordshire and the Open University in the UK and at an institution in Singapore.
I have examined PhD and DPhil theses at (at least) Bath, Birmingham, KCL, Oxford, Heriot-Watt, Loughborough, Sheffield, Southampton, Surrey, Swansea, UCL, UEA and York in the UK and at City University in Hong Kong.
I have held almost all positions of responsibility within my School, including spells in charge of all postgraduate degrees; research student admissions and progress monitoring; postgraduate development; departmental facilities; and all courses in the areas of software and computer networks. In the dim and distant past, I supervised the creation of the University's first website. I've also been involved in a number of University-wide committees, and have attended all of the institution's "leadership" courses.
I managed the accreditation by the IET and BCS of our degree programmes in March 2016, obtaining the maximum five years approval. I continue to keep an eye on the content of our degrees to make sure all the learning outcomes are being met. Being a techie type, I wrote some Python software to do most of the grunt work.
I enjoy star-gazing, and have a few friends who share the passion for amateur astronomy. I have access to a number of telescopes, including a Meade ETX-125PE and Meade Lightbridge with a whopping 16-inch mirror. If you haven't seen the moons of Jupiter or the rings of Saturn with your own eyes or looked at the star-forming regions in the Orion nebula, you're really missing out. (My favourite Messier object? M3, in Canes Venatici — think of a stellar ferrero rocher!) I also have a solar telescope, a Coronado PST, which set up in the middle of the campus on warm, sunny days and invite people to take a look through. I occasionally drag these telescopes around schools, trying to interest people in the joys of studying science subjects — do feel free to get in touch if you'd like me to visit. I have the requisite CRB checks and am a STEM ambassador.
I'm a radio amateur, M0PDF (my wife, Christine, is M0PDE), and registered instructor with RSGB. I'm one of the people who runs EARS, the Essex Amateur Radio Society, and run courses to introduce people to the pastime.
I play golf, though currently irregularly, mostly at Stoke-by-Nayland golf club. I live in hope of getting my handicap back down to 7, which is what it was when I was an 18-year-old! I also occasionally play tennis with Christine, who plays in the teams at Dedham tennis club and beats me easily (sigh).
My other main leisure interests are photography, travelling and architecture (well, looking at interesting buildings and photographing them, preferably in exotic locations). I'm a member of Alresford Camera Club — but don't confuse enthusiasm with talent: I regularly fail to do well in the club competitions. ;-) Having said that, I somwhow managed to win the Best of the Best award for the best photograph in club competitions for the 2010–11 season. In case you're interested, a small version of the winning photograph is reproduced below. It is suitably exotic: people fishing on Lake Umayo, taken with my trusty D300 at the Sillustani pre-Inca site near Lake Titicaca in Peru. Oh, and you can forget all that guff about good lighting only being available within half an hour of sunrise or sunset: this was taken shortly after noon!
I'm generally a lousy cook but occasionally manage to throw together a Thai meal. Thanks to Saint Delia, I also like to bake bread. Sadly, I developed an allergy to alcohol in my mid-20s, which makes me popular ferrying people to and from events around Christmas.
I write with a pen. You know, ink and stuff. Italic. Nib like a small spade.
My office largely Microsoft-free: I've been using a Macintosh since 2002, Linux since 1993, Emacs since 1988, and LATEX since 1984, so their idiosyncracies are second nature to me. Before coming to Essex, I was a VAX/VMS afficionado and I still regard it as the pinnacle of user-friendly operating systems. Why Microsoft-free? Well, I just don't like WYSIWYG word processors and Word is particularly difficult to become proficient with as it's full of bizarre features. In any case, how do people put up with the visual quality of Word documents? No ligatures, as far as I can see; precious little kerning; illegible maths; difficult-to-use table of contents generation; and apparently discouraging any attempts to impose document structure. As for other 'standard' tools — well, I've yet to see a non-trivial spreadsheet that didn't had a bug in it. The closest thing to a saving grace is PowerPoint but there are better alternatives out there.
I program principally Python and C. I do program in C++ and Java from time to time but always end up coming back to C. My view is that C++ is something of an abomination: it's far to easy to avoid using its object-oriented features, and the way that much of the executable source code ends up being forced to reside in the header file makes me shudder. On the other hand, Java is a nice language with generally well-conceived features (but still no operator overloading, definitely a mistake) — it's a shame its class libraries are so difficult to work with. Compared to C++ and Java, C is small, portable and easy to work in. I guess many will regard me as something of a dinosaur because, although I understand object-oriented programming and can work within its strictures, I don't particularly like them as they tend to obsfuscate the underlying algorithms, which is surely the most important part of a program. Moreover, looking at others' code, I rarely find much evidence of code encapsulation and re-use, surely one of the raisons d'etre for object-oriented programming. I seem to be able to achieve as much code re-use in ISO C, and that compiles much more quickly — and usually executes more quickly too. Or maybe it's that I can "write Fortan in any language," as the saying goes. In any case, my coding style naturally hides data, supposedly the other big win for OO languages. If I had infinite time at my disposal, I'd learn a functional programming language, as that is closer to my programming style and has the benefit of occasionally being able to be proved formally to be correct.
For about the last decade, most of my general programming has been in scripting languages as I find they are very effective for turning out working code quickly. I worked in Perl since its early days but switched to Tcl/Tk in early 2004 (I needed to use its "safe interpreter") and then to Python. I've written a numerical Python computer vision package, EVE which includes most of the vision functionality that I need. I use EVE for interactive processing, typing commands into a terminal window, as well as for short programs.
I have close ties with the Colchester Archaeology Trust (CAT) who do most of the excavations in the area; indeed, I'm a member of the Trust's management committee. (As mentioned above, Colchester was the capital of Roman Britain.) My interest in such things probably comes from being born and brought up in Hexham and the North Tyne valley. Christine and I put together the original version of the CAT web-site and, along with a couple of colleagues, I brought live video of CAT's excavations to the Internet in 1997. We believe this was the first such live excavation. More recently, Christine and I developed an online archive of CAT publications, offering things such as keyword and free-text searches and identifying reports by their locations on a map.
The photograph at the top of this page was taken by Janus van Helfteren, MD of Van Cols, a nearby company that specialises in school photographs. Janus is the stepson of the Shamim Siddiqui, a former colleague who sadly died a few months after being made a professor in my department. I liked Shamim very much and am extraordinarily chuffed with the photograph, both as a flattering likeness of me and because of the link with Shamim.
You might like to look at a VRML 1 model of Adrian's head which was kindly captured during BMVC97, held at Essex, using a C3D model 2020 system developed by the now-defunct Turing Institute in Glasgow.