Guerrilla Fleet Management
I've been doing some work on a project that uses networked cams to do guerrilla fleet management (that is, software that allows you to track someone else's vehicle fleet — for example, this could tell you where all the police cars in a given geography are). It's an interesting problem technically, and of course it's interesting in terms of power structure and surveillance culture.
Here's how it works. Cam data comes in the form of a series of timestamped images (or just a video stream, which is basically the same thing), as well as a location (which may change from frame to frame if the camera isn't mounted on a stationary point). This datastream is run through an OCR-type process which does both vehicle tracking and vehicle identification (both in terms of license, make/model, and type — police, cab, courier, and other “marked” service vehicles are easiest to see), as well as interframe and intercam comparisons to track motion and predict paths.
This data can be collected and using Markov and other statistical analysis, herd behavior, clusters, and other behavior patterns can be analyzed for efficiency, weak points, controllability, and so on. Anyway, it's an interesting tool that can empower interesting things.
Post a Comment