A near real-time, parallel and distributed adaptive object detection and retraining framework based on AdaBoost algorithm

Abstract

Object detection (e.g., face detection) using supervised learning often requires extensive training, resulting in long execution times. If the system requires retraining to accommodate a missed detection, waiting several hours or even days in some cases before the system is ready, may not be acceptable in practical implementations. This paper presents a generalized object detection framework such that the system can efficiently adapt to misclassified data and be retrained within a few minutes. The methodology developed here is based on the popular AdaBoost algorithm for object detection. To reduce the learning time in object detection, we develop a highly efficient, parallel, and distributed AdaBoost algorithm that is able to achieve a training execution time of only 1.4 seconds per feature on 25 workstations. Further, we incorporate this parallel object detection algorithm into an adaptive framework such that a much smaller, optimized training subset is used to yield high detection rates while further reducing the retraining execution time. We demonstrate the usefulness of our adaptive framework on face and car detection.