The ObjectVideo Virtual Video (OVVV) Tool generates realistic video from simulated cameras in an interactive virtual world. This tool is free and is based on a modification (aka 'mod') of Half-Life 2, a commercially available game from Valve Software.

Indoor/outdoor scenes, crowds of people, heterogeneous camera networks, UAVs and repeatable scenarios are only a few of the possibilities of virtual video ...

Our hope in distributing this tool is 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, the list goes on!).

Synthetic video is traditionally taboo in the computer vision community since it removes the artifacts that expose poor algorithms. But used appropriately for proof-of-concept, debugging, stability testing, and other important development tasks, synthetic video offers significant time and cost savings. This makes OVVV, alongside live video, an important component of your development toolbox.

Synchronized video streams from static and active cameras observing the same virtual world

Features of OVVV include:

• Freely place and configure cameras in the virtual world:
  • Generate multiple video streams from independent cameras
  • Configure intrinsic and extrinsic parameters independently for each camera
  • Manual or programmatic PTZ control allows you to simplify the testing of active vision algorithms (no more running around in the parking lot!)
  • Special camera configurations (omnicam, UAV)

Simulated omnicam and UAV video of the same crowded public area

• Automatic target ground truth generation:
  • Including target centroids, bounding boxes and pixel-wise segmentation
  • Eliminates the labour and cost of traditional manual ground truthing
  • Opens new dimensions for algorithm performance evaluation

Automatic target ground truth and realistic noise effects

• Repeatable scenarios allow you to test algorithms against:
  • Different weather conditions
  • Different types of cameras
  • Varying levels of noise/jitter, etc
  • Variable target size/motion/density, etc

Identical scenario simulated at different times of the day

• Stream real-time synthetic video through DirectShow:
  • No changes to your DirectShow-based image processing applications needed
  • Easily switch between real or synthetic video sources
  • Save video to AVI for offline processing
• Eliminate the time and cost of collecting canned test video:
  • Easily control large number of cameras (50+ multi-camera installation)
  • Easily coordinate large number of targets (dozens of people in a crowd)
  • Obtain video from logistically difficult sources (UAVs, public areas)

Using OVVV is straightforward. The tool is comprised of the following components:

• Virtual Video Mod: This mod of the commercially available game Half-Life 2 adds code to configure and control multiple virtual cameras in the game world. A copy of Half-Life 2 must be purchased and intalled (at your own expense) before this mod can be used.
• DirectShow Filter: Allows any DirectShow based application (including Windows Media Player) to retrieve video frames rendered by the game engine. If you are using DirectShow, you can start streaming real-time virtual video using this filter without any other modifications to your application. Applications that do not use DirectShow can still retrieve frames using the API described below.
• Virtual Video C Library: This C library allows applications to retrieve frames without using the DirectShow filter, and to send PTZ commands to the Half-Life 2 mod to control the camera view in real time.

The Virtual Video mod uses a socket-based network protocols to recieve PTZ commands and send video frames to client applications. Applications written in any programming language and on any platform that supports sockets can use these low-level protocols to communicate with the Half-Life 2 mod in place of the DirectShow filter and/or C Library.

For more details, see the Download/Installation and Usage pages.

ObjectVideo is the leading provider of intelligent video software for security, public safety, business intelligence gathering, process improvement and other applications. Originally founded in 1998 by world-renowned scientists and program managers from the Defense Advanced Research Projects Agency (DARPA) and headquartered in Reston, VA, ObjectVideo is a private company backed by leading venture capital partners. The company employs 100+ highly-skilled professionals and is led by an experienced management team. Nearly 40% of the company's research and development staff are PhD's, known internationally for their research in the field of computer vision. ObjectVideo products are field-proven and currently covering more perimeter miles than any intelligent video software product. The company markets and sells its products through a Video Intelligent Partner (VIP) network of OEMs, resellers, and leading security solution integrators. Currently ObjectVideo has more than 165 paid customers with deployments in 215 locations across 13 countries, providing intelligence on more than 50,000 video channels. ObjectVideo proud to support the advancement of computer vision research through ObjectVideo's Virtual Video Tool.