How Gait Recognition and Neighbor Cameras Aid Identification (Full Transcript)
[00:00:00] Speaker 1: The video shows the person walking away to find something. I guess that's when he grabs the flowers to obscure the camera. And so in that image, though, you do see the person move. You see how they turn around, how they lean, how they walk, carrying the backpack. You know, I guess it would seem, Christian, that if you knew this person, some of that without even realizing why, would be familiar to you?
[00:00:26] Speaker 2: Oh, absolutely. I mean, along with the facial recognition technology, there's gait recognition technology, which is just as accurate, because the way we walk and the way we move is as much a fingerprint for us as our fingerprints, as our faces. And what's going to be interesting here is figuring out a way to take all of these features and use them to guide the mining of other videos. And when I mention the neighbors, it's like the reality is that there will be people in the neighborhood who also have cameras. And those cameras may have a different subscription service, and there may be more data on those cameras. And getting to that might actually get to more information about the movement, not in the house, but outside the house.
- Gait recognition can be as distinctive and useful as facial recognition for identifying individuals.
- Subtle movement cues (turning, leaning, walking style) may be recognizable to acquaintances.
- Combining multiple features can help guide the search/mining of additional videos.
- Neighboring cameras may provide crucial external movement data, potentially with richer metadata depending on the service.
- Cross-service access to camera footage can be important for reconstructing events outside a home.