Robust Semi-Supervised Traffic Sign Recognition via Self-Training and Weakly-Supervised Learning.

Obed Tettey Nartey,Sarpong Kwadwo Asare,Lady Nadia Frempong,Jinzhao Wu,Guowu Yang

doi:10.3390/s20092684

Abstract

Traffic sign recognition is a classification problem that poses challenges for computer vision and machine learning algorithms. Although both computer vision and machine learning techniques have constantly been improved to solve this problem, the sudden rise in the number of unlabeled traffic signs has become even more challenging. Large data collation and labeling are tedious and expensive tasks that demand much time, expert knowledge, and fiscal resources to satisfy the hunger of deep neural networks. Aside from that, the problem of having unbalanced data also poses a greater challenge to computer vision and machine learning algorithms to achieve better performance. These problems raise the need to develop algorithms that can fully exploit a large amount of unlabeled data, use a small amount of labeled samples, and be robust to data imbalance to build an efficient and high-quality classifier. In this work, we propose a novel semi-supervised classification technique that is robust to small and unbalanced data. The framework integrates weakly-supervised learning and self-training with self-paced learning to generate attention maps to augment the training set and utilizes a novel pseudo-label generation and selection algorithm to generate and select pseudo-labeled samples. The method improves the performance by: (1) normalizing the class-wise confidence levels to prevent the model from ignoring hard-to-learn samples, thereby solving the imbalanced data problem; (2) jointly learning a model and optimizing pseudo-labels generated on unlabeled data; and (3) enlarging the training set to satisfy the hunger of deep learning models. Extensive evaluations on two public traffic sign recognition datasets demonstrate the effectiveness of the proposed technique and provide a potential solution for practical applications.

Highlights

Traffic signs provide reliable safety precautions and guiding information to road users on highways, motorways, urban surroundings, and the sort
We propose a novel ROSST framework that takes into consideration the challenges of imbalanced datasets by utilizing weakly-supervised learning to generate attention maps representing the spatial distributions of an object’s parts, to extract local features, and via self-paced learning, to solve a small sample problem using the traffic sign recognition problem
We proposed a robust semi-supervised learning (ROSST)

Summary

Introduction

Traffic signs provide reliable safety precautions and guiding information to road users on highways, motorways, urban surroundings, and the sort. Most methods that have been deployed for traffic sign recognition, be it traditional computer vision methods or advanced ones, have used a supervised learning approach. Classical supervised learning demands all samples to be well annotated before a good model can be built, which is a major drawback when factors such as labeling cost, time, and demand for expertise knowledge are considered. To reduce the labeling cost and make use of both labeled and unlabeled data, a semi-supervised learning technique is used. Semi-supervised learning is an approach that automatically assigns a class to unlabeled samples by relying on its capabilities of predicting labels correctly and through training, extending its knowledge on the predictions learned and/or its competence in classifying [1]

Objectives

Methods

Results

Conclusion