ClickBAIT-v2: Training an Object Detector in Real-Time

TL;DR

This paper introduces ClickBAIT-v2, a real-time object detector training system that leverages weak supervision and object tracking to minimize human effort and enable online learning from live video streams.

Contribution

It presents a novel system for online training of object detectors using weak supervision and object tracking, reducing human annotation effort in real-time scenarios.

Findings

Object tracking increases training efficiency 3-4 times.

Weakly-supervised single-point clicks can generate effective bounding box annotations.

The system enables real-time object detection training with minimal human input.

Abstract

Modern deep convolutional neural networks (CNNs) for image classification and object detection are often trained offline on large static datasets. Some applications, however, will require training in real-time on live video streams with a human-in-the-loop. We refer to this class of problem as time-ordered online training (ToOT). These problems will require a consideration of not only the quantity of incoming training data, but the human effort required to annotate and use it. We demonstrate and evaluate a system tailored to training an object detector on a live video stream with minimal input from a human operator. We show that we can obtain bounding box annotation from weakly-supervised single-point clicks through interactive segmentation. Furthermore, by exploiting the time-ordered nature of the video stream through object tracking, we can increase the average training benefit of human interactions by 3-4 times.