1. Computer Vision: Three person tracking and localization

    01:13

    from CoVar / Added

    6 Plays / / 0 Comments

    Example video showing approximate tracking of three people in a scene. Occasional missed-localizations are caused by inter-person aliasing. When people become too close together, their individual tracks are temporarily lost. Overall, the right number of people and their approximate locations are estimated well. Future work should include positioning cameras from a bird's-eye-view to help with inter-person occlusion and aliasing.

    + More details
    • Computer Vision: Two person tracking with ball throwing

      00:34

      from CoVar / Added

      5 Plays / / 0 Comments

      Example of tracking the locations of two people as they play with a football, and one juggles (poorly). 6 cameras are used to localize the two people as they move around the scene.

      + More details
      • Computer Vision: Person tracking in 6 cameras

        00:58

        from CoVar / Added

        2 Plays / / 0 Comments

        Example of tracking the locations of two people as they move around a scene and carry on a conversation. 6 cameras are used to localize the two people as they move around the scene.

        + More details
        • Computer Vision: Single person tracking

          00:32

          from CoVar / Added

          2 Plays / / 0 Comments

          Example of tracking a persons location as they move around in a scene. Here, 6 cameras are used for robustness to temporary occlusions and missing detections.

          + More details
          • Computer Vision: Personnel video monitoring - person enter, exit

            00:22

            from CoVar / Added

            22 Plays / / 0 Comments

            An example of personnel tracking with multiple cameras. As a person enters the scene, they are detected in individual cameras, and their location is estimated in the actual 3-D scene. As they move, their position is updated. The person picks up a tool, and exits the scene and is tracked throughout.

            + More details
            • Computer Vision: Real-time demonstration of personnel video monitoring

              00:51

              from CoVar / Added

              26 Plays / / 0 Comments

              Example real-time demonstration of personnel video monitoring. Personnel are detected in each camera independently, and person locations are calculated using triangulation. The system operates in C++ in parallel on four computers; communications are accomplished using Google protocol buffers and UDP over ethernet. OpenCV is leveraged for the cv::Mat objects, but all detection, tracking and localization is performed using internal CoVar technologies.

              + More details
              • Computer Vision: Personnel video monitoring - under the hood

                00:41

                from CoVar / Added

                14 Plays / / 0 Comments

                Example under-the-hood visualization of real-time personnel video monitoring. Frames on the left show persons in four cameras positioned in four corners of a room. Person detections in each camera are shown in blue boxes, and aggregated (across-camera) person detections are shown in the red ovals projected on the ground-plane. In the central image, a spatial confidence map shows how the likelihood of a person changes and evolves spatially. When people are standing in specific locations, "ghosts" appear due to bearings-only information aliasing. In the right image, the results of persistent object tracking are shown; the actual people are accurately tracked as they move in the scene.

                + More details

                What are Tags?

                Tags

                Tags are keywords that describe videos. For example, a video of your Hawaiian vacation might be tagged with "Hawaii," "beach," "surfing," and "sunburn."