ObjectVideo Virtual Video SDK is a tool that generates realistic video from simulated cameras in a virtual environment. It is designed to facilitate computer vision research in areas that cannot use prerecorded video or require large amounts of ground truth data (e.g. multi-camera installations, public spaces, etc.). Features include generating multiple video streams from independent cameras, eliminating the need for manual ground truth, and providing new dimensions for performance evaluation. It also allows users to easily switch between real and virtual video.
The ObjectVideo Virtual Video generates realistic video from simulated cameras in an interactive virtual world. This tool aims to stimulate computer vision research in areas that cannot rely on canned video (eg. active tracking) or when large quantities of ground truthed video is unavailable or impractical (multi-camera installations, public spaces, etc.).
Features:
- Generates multiple video streams from independent cameras.
- Eliminates the labour and cost of traditional manual ground truthing.
- Opens new dimensions for algorithm performance evaluation.
- Easily switch between real or synthetic video sources.
- Obtains video from logistically difficult sources (UAVs, public areas).
Comments