Semi-Automatic Annotation For Visual Object Tracking

TL;DR

This paper introduces a semi-automatic annotation method for visual object tracking that leverages temporal information and human verification to significantly reduce manual annotation effort.

Contribution

It presents a novel incremental learning approach combining tracking-by-detection and human-in-the-loop evaluation to improve annotation efficiency.

Findings

Annotation workload reduced by up to 96%

Effective use of MHT for false-positive reduction

Iterative training improves detection accuracy

Abstract

We propose a semi-automatic bounding box annotation method for visual object tracking by utilizing temporal information with a tracking-by-detection approach. For detection, we use an off-the-shelf object detector which is trained iteratively with the annotations generated by the proposed method, and we perform object detection on each frame independently. We employ Multiple Hypothesis Tracking (MHT) to exploit temporal information and to reduce the number of false-positives which makes it possible to use lower objectness thresholds for detection to increase recall. The tracklets formed by MHT are evaluated by human operators to enlarge the training set. This novel incremental learning approach helps to perform annotation iteratively. The experiments performed on AUTH Multidrone Dataset reveal that the annotation workload can be reduced up to 96% by the proposed approach.