Analysis and a Solution of Momentarily Missed Detection for Anchor-based Object Detectors

TL;DR

This paper analyzes the cause of momentary missed detections in anchor-based object detectors in videos and proposes a training improvement to mitigate this issue.

Contribution

It identifies the boundary behavior of anchor-based detectors as a key cause of missed detections and suggests a new sampling method during training to address this.

Findings

Most missed detections are due to boundary anchor box issues.

Improved positive sample selection reduces momentary missed detections.

Analysis clarifies causes beyond common factors like occlusion or motion blur.

Abstract

The employment of convolutional neural networks has led to significant performance improvement on the task of object detection. However, when applying existing detectors to continuous frames in a video, we often encounter momentary miss-detection of objects, that is, objects are undetected exceptionally at a few frames, although they are correctly detected at all other frames. In this paper, we analyze the mechanism of how such miss-detection occurs. For the most popular class of detectors that are based on anchor boxes, we show the followings: i) besides apparent causes such as motion blur, occlusions, background clutters, etc., the majority of remaining miss-detection can be explained by an improper behavior of the detectors at boundaries of the anchor boxes; and ii) this can be rectified by improving the way of choosing positive samples from candidate anchor boxes when training the detectors.