# 4-D Scene Alignment in Surveillance Video

## TL;DR

This paper introduces an automatic 4-D scene alignment method for surveillance videos that combines CNN-based camera pose estimation with pedestrian observations, enabling robust activity detection without explicit tracking.

## Contribution

It presents a novel automatic calibration approach that estimates 4-D scene geometry without tracking or explicit detection of head and feet, improving robustness.

## Key findings

- Robust camera calibration without tracking or explicit detection.
- Effective handling of height variations and camera errors.
- Enhances activity detection in surveillance videos.

## Abstract

Designing robust activity detectors for fixed camera surveillance video requires knowledge of the 3-D scene. This paper presents an automatic camera calibration process that provides a mechanism to reason about the spatial proximity between objects at different times. It combines a CNN-based camera pose estimator with a vertical scale provided by pedestrian observations to establish the 4-D scene geometry. Unlike some previous methods, the people do not need to be tracked nor do the head and feet need to be explicitly detected. It is robust to individual height variations and camera parameter estimation errors.