Researchers around the world are looking into wearable video camera to gather live data associated with movement of the person, utilize the data as memory aid tool or diagnosis of autism. The camera can help patients immediately locating their lost keys, remembering old faces and ease other functions. Researchers at the University of Bordeaux have solved a tough problem of categorizing the lifestream data collected by body-mounted camera to detect dementia.
Challenge with life stream data
The major challenge with a wearable video camera is that it captures each and every moment of person’s life. The biggest challenge is, how would you make sense of this huge data set?
- In movies and Tv serials scene change can be detected easily as it usually coincides with a change in the camera’s perspective
- In live stream, perspective is always same and images are complicated by movement, too much light etc, which makes the task of change recognition harder.
What is proposed solution?
Svebor Karaman et amis at the University of Bordeaux in France have found a new way of categorizing daily activities in footage taken from a shoulder-mounted camera. This is the proposed solution:
- scene is as a sequence of frames in which the camera is relatively still, which can be easily determined by measuring trajectory of the corners of the image
- each sequence is then categorized according to the colors present in the frames, which remain relatively constant even when the individual frames are blurred or dim.
- Finally, scenes are...
manually labeled as “moving in the kitchen” or “moving in home office”
This categorization is expected to result in a reasonably accurate picture of the activities that an ordinary person caries out on a daily basis.
The motivating factor for Karaman and co is to find an objective way of studying the patterns of behavior of people with dementia. As doctors rely on information regarding person’s behavior from friends or care takers, which can be biased. Karaman and co say “that this kind of data can be an important tool in evaluating the onset of dementia and the way it is advancing.”
This is a significant development to utilize data from wearable cam. Karaman and co said that there is a lot of work to refine the technique, so that the program can recognize things like kitchens, office, etc. Well good luck in advancing the technique and hope to see it utilized in benefit of dementia patients.