Systems and Methods for Multi-Pass Adaptive People Counting

A method of operating a video processing system to count objects in video, comprising:

• analyzing at least a segment of the video in accordance with at least one initial parameter to produce an initial count;
• identifying a subsequent parameter with which to analyze the video based on at least the initial count;
• analyzing the segment of the video in accordance with at least the subsequent parameter to produce a subsequent count;
• and
• venerating a plurality of trajectories corresponding to movement of the objects in the video and identifying the subsequent parameter with which to analyze the video based on the trajectories in addition to at least the initial count.

A non-transitory computer readable media having stored thereon program instructions for operating a video processing system to count objects in video, wherein the program instructions, when executed by the video processing system, direct the video processing system to at least:

• analyze at least a segment of the video in accordance with at least one initial parameter to produce an initial count;
• identify a subsequent parameter with which to analyze the video based on at least the initial count;
• analyze the segment of the video in accordance with at least the subsequent parameter to produce a subsequent count;
• and
• generating a plurality of trajectories corresponding to movement of the objects in the video and identifying:
• the subsequent parameter with which to analyze the video based on the trajectories in addition to at least the initial count.

A method of counting people in a video using multiple passes, comprising:

• counting people in a segment of video with a video processing system configured with a first set of parameters to produce a first output;
• determining a second set of parameters based on at least the first output;
• and,
• counting people in the segment of video with the video processing system configured with the second set of parameters to produce a second output;
• wherein the second set of parameters is chosen from a group of sets of parameters that include sets of parameters optimized for low congestion scenes, high congestion scenes, and standard congestion scenes.