Hi Alice, Have you tried post-processing the video to still images? Then it's a simple, albeit rough, way to scan your data set quickly.
With the Cornell Lab of Ornithology's NestCams project we also capture still images and post-process 24 hours worth of images (~5000 images) into <2 min condensed image clips. YouTube channel here: http://www.youtube.com/user/CLOCitizenScience and NestCams project is here: http://nestcams.org In terms of reviewing and categorizing the image archive, we have developed CamClickr, a crowdsourcing-Citizen Science application where participants review our still image archive and tag the images. http://camclickr.org You might also take a look at Collaborative Observatories for Natural Environments (CONE). One of their systems actually tags motion of interest and condenses it into a time lapse clip for later review. http://www.c-o-n-e.org/ Aside from more complex computer vision/machine learning projects - I don't think the technology is there yet. Kitty Project Assistant Ready for a New Opportunity NestCams.org & CamClickr.org Cornell Lab of Ornithology
