Engineer builds DIY ALPR rig to track police cars; police visit his home
Canadian software engineer and YouTuber Anthony Sistilli said two officers from Ontario's Peel Regional Police visited his home on September 20 to ask why he built a surveillance device aimed at tracking police vehicles. The prototype uses an roughly $180 camera and automated license plate recognition (ALPR): image recognition first decides whether a vehicle is a police car, and only then tries to read its plate, with AI coding tools used in development and the pipeline split into two stages to cut the compute load. Sistilli said the results were limited — a one-hour test produced only two police-car images, with plates hard to read — making it a proof-of-concept rather than a working tracking system, while police raised concerns that continuously collecting where and when patrol cars appear could reveal officers' routines.
It shows how consumer hardware plus off-the-shelf AI recognition and coding tools are lowering the barrier for individuals to build vehicle-surveillance systems, widening the privacy debate beyond large institutions.
加拿大软件工程师兼YouTube博主安东尼·西斯蒂利（Anthony Sistilli）近日公开讲述了这段经历。他表示，9月20日，两名来自加拿大安大略省皮尔区地区警察局（Peel Regional Police）的警员来到他家中，询问他为什么要开发一套专门追踪警车的监控设备。
西斯蒂利此前发布了多段视频，展示这套设备的开发过程和实际测试效果。他将自己的项目形容为一套针对Flock监控系统的个人监控设备，其核心理念是利用警方日常使用的技术，反过来观察警察车辆的活动。
这套设备采用自动车牌识别技术，也就是ALPR（Automated License Plate Recognition）。这项技术通常通过摄像头拍摄道路上的车辆，再利用图像识别算法提取车牌号码，并将车牌、拍摄时间和地点等信息保存下来，供后续查询。
与普通的车牌识别设备不同，西斯蒂利设计的系统并不打算记录道路上所有车辆的信息，而是首先识别画面中的车辆是否为警车。只有当系统认为目标车辆属于警用车辆时，才会进一步尝试读取其车牌号码。