Incremental Deep Learning for Robust Object Detection in Unknown Cluttered Environments

• Published in: IEEE access 2018, Vol.6, p.61748-61760
• Main Authors: Shin, Dong Kyun, Ahmed, Minhaz Uddin, Rhee, Phill Kyu
• Format: Article
• Language: English
• Subjects: active learning; Artificial neural networks; Changing environments; convolutional neural network; Deep learning; Detectors; Image quality; incremental deep learning; Knowledge management; Labeling; Labels; Machine learning; Moving object recognition; Neural networks; Noise measurement; Object detection; Sampling methods; semi-supervised learning; Semisupervised learning; Streaming media; Training
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2018.2875720

Object detection in streaming images is a major step in different detection-based applications, such as object tracking, action recognition, robot navigation, and visual surveillance applications. In most cases, image quality is noisy and biased, and as a result, the data distributions are disturbed and imbalanced. Most object detection approaches, such as the faster region-based convolutional neural network (RCNN), single shot multibox detector with 300Œ300 inputs (SSD300), and you only look once version 2 (YOLOv2), rely on simple sampling without considering distortions and noise under real-world changing environments, despite poor object labeling. In this paper, we propose an incremental active semi-supervised learning (IASSL) technology for unseen object detection. It combines batch-based active learning (AL) and bin-based semi-supervised learning (SSL) to leverage the strong points of AL's exploration and SSL's exploitation capabilities. A collaborative sampling method is also adopted to measure the uncertainty and diversity of AL and the confidence in SSL. Batch-based AL allows us to select more informative, confident, and representative samples with low cost. Bin-based SSL divides streaming image samples into several bins, and each bin repeatedly transfers the discriminative knowledge of convolutional neural network deep learning to the next bin until the performance criterion is reached. The IASSL can overcome noisy and biased labels in unknown, cluttered data distributions. We obtain superior performance, compared with the state-of-the-art technologies, such as Faster RCNN, SSD300, and YOLOv2.