In October 2009 Microsoft released an experimental desktop application called Pivot which allowed for the exploration of large collections of images. In 2010 it was showcased to much fanfair by Gary Flake, a Technical Fellow at Microsoft (now the CTO Search and Data Science at salesforce.com), who presented it as a novel way for looking at data in a TED talk.
“Gary Flake demos Pivot, a new way to browse and arrange massive amounts of images and data online. Built on breakthrough Seadragon technology, it enables spectacular zooms in and out of web databases, and the discovery of patterns and links invisible in standard web browsing”
Soon after in June 2010 Microsoft then released a Silverlight control based on Pivot called PivotViewer, which allowed .Net developers to integrate it into their Silverlight and web applications.
I remember when I first saw PivotViewer and being blown away at the possibilities that it provided – an entirely visual way of looking at and exploring data. I was a Business Intelligence consultant at the time and It made working with data fun and engaging, allowing me to create whole new experiences for my clients. I instantly become familiar with all aspects of how to create collections and host them on the web.
I was further inspired by the 10-part series “Adventures with PivotViewer” by xpert360, where they explored how they customised various parts of the PivotViewer control to add additional functionality. I started experimenting myself and created a dynamic collection of photos that queried the Flicker API in real time. I then further extended PivotViewer to include a Bing maps view which made it possible to view where each photo was taken. As I was getting deeper into working with PivotViewer I soon realised that for various reasons Silverlight as a platform had no future. The solution was to develop an alternative that was cross platform, utilised web standards and customisable. To that end I stated work on the HTML5 PivotViewer and in 2012 launched it to the world with a blog post titled Addressing the elephant in the room: The HTML5 PivotViewer.
At the time I was (and still am) by no means a huge force in the blogging world, but to date, that one post has managed to eclipse all my other posts combined. Looking at the data I knew I was on to something. It was through this one post that I came across many others that also shared my love for PivotViewer and were using it for anything from online shopping to cataloguing paint samples. It was through this humble beginning that I met two people that were to play a significant role in my future. Stephen Taylor, the Head of Computational Biology of the Weatherall Institute of Molecular Medicine, University of Oxford was using PivotViewer for high-throughput microscopy and wanted to use the HTML5 version for future research. Over a period of 18 months we both worked to further refine the HTML5 PivotViewer and eventually co-authored a paper titled “HTML5 PivotViewer: high-throughput visualization and querying of image data on the web” which was published in Bioinformatics, May 2014.
At the same time in 2012 I met Samuel Conway, the Managing Director of a SharePoint consultancy, Coritsu Group in Adelaide, Australia. Samuel had seen a PivotViewer implementation that I had done a mutual client and had a strong vision for how it could be used for document management within SharePoint. I started working with Sam on various projects and eventually came on board at Coritsu.
After the paper was published the relationship with Oxford University continued to grow. Working with the University, over an 8 month period, Coritsu Group was able to build the next version of PivotViewer in which Zegami was born. The success of the project and the subsequent development led to the formation of Zegami Limited a spinout company of the University of Oxford on the 1st of Feb 2016.
Zegami is a significant step forward for PivotViewer and the way that images are handled. We’ve taken something that was a tool for developers to build applications with and turned it into a packaged, deployable solution available on any device. Zegami is not only more feature complete, it takes care of generating and managing very large collections of images (no more Excel plugin!). In addition:
- Display tens of thousands of images
- Grouping, Map and Scatterplot Filters
- Ease of collection generation with the administration portal
- Security model to control who can access a collection
- Plugin API that allows for the development of new functions and features
- Manage facets from within the management portal (no CXML!)
- Visual image search
So what does the future hold for Zegami?
As an Oxford spinout the pressure is now on to not only commercialise Zegami but also take it to the next level of functionality and the team at Zegami have no doubt where that needs to be. Image processing techniques have advanced significantly over the past few years and we can use them in a way that was not possible when PivotViewer was launched in 2010. We are integrating more and more machine learning and deep learning to put advanced image processing, visual search and classification techniques into the hands of everyday users. We will continue to develop our API to allow developers to enhance and customise Zegami to meet their needs.
PivotViewer vs. Zegami – there is no comparison. Let Zegami help you see what you have been missing….