BBC Video About The SenseCam

This video is a bit old but I just found it so it is new to me: James May of the BBC show about “Big Ideas” where he takes a quick gander at Microsoft’s SenseCam. James wore the SenseCam he was loaned for just a weekend and immediately hit the issue that everyone who wears a SenseCam already has, the sheer amount of data captured by the device.

How do you make sense (a pun) of it all? It is interesting to see that the software that the Microsoft researchers have come up with is no more sophisticated than the Python software I have managed to develop to do the same kinds of manipulations, i.e. locate significant events and in a semi-autonomous way tag them.

Over the past week or so I have been experimenting with SURF and SIFT using OpenCV and Python to automatically determine places I have been to before. I can do this with a GPS but it would be nice to have an automated process, which is about 70% feature complete right now, that can recognise rooms I have been in at the office, or at home, or other locations that I frequent, and automatically group them together.

I am also working on using OpenCV to automatically recognise human faces and group images together, automatically tagging those people that I know and indicating people that are not tagged.

I am still making use of Windows Live Photo Gallery, simply because it offers some very fast image browsing and tagging functionality, along with PhotoSynth, but I have begun to use it less and less as my own application develops new features. With SURF analysis I have an almost complete PhotoSynth clone that can create a 3D scene from all of my SenseCam images.

I am wondering how Alan, the researcher working on software, is able to automatically determine significant events in a day. Currently I am wrestling with this problem by looking for a gathering of human faces, significant light changes in the environment or spending time within a small geographic region determined by geo-location that is outside of my normal pattern or existence. But I have yet to fathom how the Microsoft software does it.

 

 

It is good to see the popular press taking an interest in these devices, but I still fear that they are focusing on the wrong thing, turning their attention to the “man jewellery” rather than what happens after the images have been captured.