As I’ve been scanning in my photos, negatives and slides I have also been playing with various software options for saving, finding and sorting. One of the big issues with having a huge virtual stack of photos is being able to figure out the who, what and when specific to the subject matter ! For the most part, I’m not bothering to scan photos that do not have people in them, but of course my existing collection of photos taken since about 2000 has been digital and so does include lots of scenic (at least to me !) shots. This then provides an excellent opportunity to try out one of the facial recognition application options recently offered by various suppliers. There are a number to choose from depending on the platform you are running – for a good overview take a look at this.
Facial recognition capability grew out of the security industry, and it has many applications in both security as well as industry and consumer. Social network sites like Facebook currently have a type of recognition software built-in that picks out faces in photos you upload to the site, and which you can then tag with names, but the current version at least does not search through your uploaded photo collection and attempt to automatically tag other photos having the image of the same person. This sort of capability is available in the iPhoto application for Mac, Live Photo Gallery from Microsoft, and also for free in Picasa 3.6 offered by Google. While normally it would be true to say that you only ever get what you pay for, in this case free isn’t too bad at all !
I started with a test load of about 5,000 photos and left Picasa to crank through them for a while – the CPU loading was never very high (perhaps averaging about 15% on my machine) but it did take quite a while to complete as the photo store is on the server elsewhere in the house, and in this case, running over the network is slower than running on a local store. In fact, I don’t know when it actually completed as I started it off and then went to bed, but I can say that in the morning it was ready and waiting. During the night, it had picked out a range of faces and then the process is you tag these with suitable naming. Picasa then starts doing comparisons on its face collection and groups the photos by named face.
I have to admit, I was blown away by how well this worked, although there are a few things to be aware of. First of all, photos not oriented square to the camera (so not upside down or tilted at 90 degrees to the frame) are not matched. This means that to ensure your entire collection is scanned for faces, and then tagged as faces are identified, you must first go through your collection and flip them round to the ‘correct’ orientation. While I did not have any problem with ‘non-human’ objects being tagged as faces, I did get all the facial images that comprise Mount Rushmore carefully tagged which I suppose is a credit to the sculptor. Picasa also tagged facial images in background photos that appeared in photos, and also images in paintings and sculptures that were in the frame. I actually found this to be quite impressive, as in some cases the quality of these background images was not good. What I hadn’t considered though was that photos taken at our son’s graduations, each having literally hundreds of faces in the frame, could possibly generate so many face images which then had to be tagged as ‘ignore’ !
The other interesting thing was how fast it seemed to learn – initially Picasa was easily confused by family members each having very similar facial features, but once there were a few dozen validated images, it seemed to get remarkably accurate. That’s the rub though – it does take a fair bit of hands on, and so you do need to be prepared to invest some effort, but I think its worth it.
I am now in the process of cataloguing a folder containing about 30,000 photos, and Picasa has so far found about 17,000 faces which I’m in the process of sorting into about 200 ‘buckets’, each being a family member, past and present day on the three continents we are spread across. For our sons for example – it is fascinating to be able to easily pull up a library of thumbnails showing each of them as they grew up, and see how they have changed over the past 20 odd years. If you play through the images quickly, it’s almost like watching them morph from being children into adults, and is slightly hypnotic ! It is also not difficult to able to call up a portfolio of those photos for display on our TV using our Asus O!Play device, or to be used as screen savers or as desktop background images on the computers we use.
Bottom line is that software such as this turns photos from being a static resource, to being a dynamic one, and as such is pretty special. Recommended !