Early access to Vision SDK

We are about to launch the Public Beta of the Vision SDK — making it available for all developers. We’re letting a few more people in before we go live — just email vision@mapbox.com, let me know why you are excited and I’ll add you to this list. The first 100 responses will get added.

The Vision SDK processes imagery directly on mobile and embedded devices, so any connected camera becomes a second set of eyes for your car. The neural networks run AI-powered image processing at super low power, so it runs on phones and other IoT devices.

We just added 3 key new features:

1. Safety alerts
2. WiFi dashcam tethering for to users that want to keep their phones close-at-hand
3. Linux support allowing automotive and other embedded developers to run Vision on custom hardware

Sneak peek of safety alerts

We detect pedestrians, cyclists, and other vehicles in a driver’s path. Developers use these detections to send custom alerts to drivers using the Vision Safety module. For example, use sign detection to send drivers alerts when they exceed the speed limit, or distance measurements to warn a driver about tailgating. The safety features of the Vision SDK are especially useful for fleet companies — fleet operators can use Vision to detect road incidents and automatically document them with supporting images or video.

Layer augmented reality on top of your navigation experience

Augmented reality tools in the Vision AR module project 3D turn-by-turn directions onto the road ahead. In the new release, you can place other objects in the environment, such as information on points of interest (POI), parking availability, or even custom data sets. Place these objects based on geocoded information (such as POI pins), or by using detection and segmentation information from the Vision SDK.

More on mapbox.com/vision. For mobile platforms (iOS or Android), Vision is free for up to 100 monthly active users, so you can start prototyping right away. Full pricing information will be provided at launch.