Grouping Crowd-Sourced Mobile Videos for Cross-Camera Tracking

Abstract

Public adoption of camera-equipped mobile phones has given the average observer of an event the ability to capture their perspective and upload the video for online viewing (e.g. YouTube). When traditional wide-area surveillance systems fail to capture an area or time of interest, crowd-sourced videos can provide the information needed for event reconstruction. This paper presents the first end-to-end method for automatic cross-camera tracking from crowd-sourced mobile video data.Our processing (1) sorts videos into overlapping space-time groups, (2) finds the inter-camera relationships from objects within each view, and (3) provides an end user with multiple stabilized views of tracked objects. We demonstrate the system's effectiveness on a real dataset collected from YouTube.

Automatic Non-Linear Video Editing Transfer

“…We represent the camera motion as a reprojection of video frames, denoted as a projective transformation H ref . Our method captures the camera motion through frame-to-frame keypoint matching with a RANSAC based homography solver [9,10] (see Figure 3). We reject keypoints on a face or body to remove foreground objects from the homography estimation of the camera's actual motion.…”

Proceedings of the 22nd Conference on Computer Aided Architectural Design Research in Asia (CAADRIA)

“…Some research realized marker-less AR using Structure from Motion (SfM) technique, which can reconstruct 3D shapes from photographs of plural viewpoints (Dellaert et al 2000). However, AR systems using SfM (Frey et al 2013;Sato et al 2016) have specific limitations. For example, there is a time lag between the process of point cloud reconstruction and the process of system use.…”