A Small Network MicronNet-BF of Traffic Sign Classification

Abstract

One of a very significant computer vision task in many real-world applications is traffic sign recognition. With the development of deep neural networks, state-of-art performance traffic sign recognition has been provided in recent five years. Getting very high accuracy in object classification is not a dream any more. However, one of the key challenges is becoming making the deep neural network suitable for an embedded system. As a result, a small neural network with as less parameters as possible and high accuracy needs to be explored. In this paper, the MicronNet which is a small but powerful convolutional neural network is improved by batch normalization and factorization, and the proposed MicronNet-BN-Factorization (MicronNet-BF) takes advantages about reducing parameters and improving accuracy. The effect of image brightness is reduced for feature recognition by the elimination of mean and variance of each input layer in MicronNet via BN. A lower number of parameters are realized with the replacement of convolutional layers in MicronNet, which is the inspiration of factorization. In addition, data augmentation is also been changed to get higher accuracy. Most important, the experiment shows that the accuracy of MicronNet-BF is 99.383% on German traffic sign recognition benchmark (GTSRB) which is much higher than the original MicronNet (98.9%), and the most influence factor is batch normalization after the confirmation of orthogonal experimental. Furthermore, the handsome training efficiency and generality of MicronNet-BF indicate the wide application in embedded scenarios.

RF-CNNS: Thin Deep Learning Networks For Accelerating Traffic Signs Recognition

“…For the last presented approaches, the Binarized Neural Network used by Postovan and Erascu (2023) achieves an accuracy of 88.17% (for 23 classes of BTSCD). For MicronNet-BF (Fang et al, 2022), which is based on factorization and batch normalization, it achieves 82.122%, while using a relatively small number of parameters (0.44 million), with GPU.…”

“…For this specific reason, many researchers adopt instead DL based approaches, to ensure an automatic extraction of high-level features from raw input data. We find hence that many deep neural networks are proposed, in the last decade, to ensure traffic signs recognition (Jurišić et al, 2015;Arcos-García et al, 2018;Li et al, 2019;Mehta et al, 2019;Zaibi et al, 2021;Fang et al, 2022), etc.…”

Recent Advances in Traffic Sign Recognition: Approaches and Datasets

“…Fang et al, (2022) [ 32 ] presented a method for traffic sign classification using MicronNet-BN-Factorization (MicronNet-BF). MicronNet is a small deep neural network designed for use in embedded devices, and MicronNet-BF improved its accuracy by integrating it with batch normalization and factorization.…”