Context R-CNN: Long Term Temporal Context for Per-Camera Object Detection

TL;DR

This paper introduces Context R-CNN, an attention-based model leveraging long-term temporal context from unlabeled frames to enhance object detection in static cameras with irregular sampling, showing significant performance improvements.

Contribution

The paper presents a novel attention-based approach that uses a per-camera long-term memory bank to incorporate extended temporal context for improved detection accuracy.

Findings

Context R-CNN improves detection performance over baselines.

Longer temporal context horizons lead to better results.

Significant mAP gains on camera trap and traffic camera datasets.

Abstract

In static monitoring cameras, useful contextual information can stretch far beyond the few seconds typical video understanding models might see: subjects may exhibit similar behavior over multiple days, and background objects remain static. Due to power and storage constraints, sampling frequencies are low, often no faster than one frame per second, and sometimes are irregular due to the use of a motion trigger. In order to perform well in this setting, models must be robust to irregular sampling rates. In this paper we propose a method that leverages temporal context from the unlabeled frames of a novel camera to improve performance at that camera. Specifically, we propose an attention-based approach that allows our model, Context R-CNN, to index into a long term memory bank constructed on a per-camera basis and aggregate contextual features from other frames to boost object detection performance on the current frame. We apply Context R-CNN to two settings: (1) species detection using camera traps, and (2) vehicle detection in traffic cameras, showing in both settings that Context R-CNN leads to performance gains over strong baselines. Moreover, we show that increasing the contextual time horizon leads to improved results. When applied to camera trap data from the Snapshot Serengeti dataset, Context R-CNN with context from up to a month of images outperforms a single-frame baseline by 17.9% mAP, and outperforms S3D (a 3d convolution based baseline) by 11.2% mAP.