Why Accuracy Is Not Enough: The Need for Consistency in Object Detection

TL;DR

This paper highlights the importance of consistency in object detection, demonstrating that current models vary in predictions due to small distortions, and proposes a new video-based measurement method to evaluate and improve this aspect.

Contribution

It introduces a novel method to measure detection consistency over time using real video frames and shows how image distortion corrections can enhance this consistency.

Findings

Detection consistency ranges from 83.2% to 97.1% across datasets.

Applying image distortion corrections improves consistency by up to 5.1%.

Consistency issues are not fully addressed by accuracy metrics alone.

Abstract

Object detectors are vital to many modern computer vision applications. However, even state-of-the-art object detectors are not perfect. On two images that look similar to human eyes, the same detector can make different predictions because of small image distortions like camera sensor noise and lighting changes. This problem is called inconsistency. Existing accuracy metrics do not properly account for inconsistency, and similar work in this area only targets improvements on artificial image distortions. Therefore, we propose a method to use non-artificial video frames to measure object detection consistency over time, across frames. Using this method, we show that the consistency of modern object detectors ranges from 83.2% to 97.1% on different video datasets from the Multiple Object Tracking Challenge. We conclude by showing that applying image distortion corrections like .WEBP Image Compression and Unsharp Masking can improve consistency by as much as 5.1%, with no loss in accuracy.