10 Hits in 3.7 sec

MR-CNN: A Multi-scale Region-based Convolutional Neural Network for Small Traffic Sign Recognition

Zhigang Liu, Juan Du, Feng Tian, Jiazheng Wen
2019 IEEE Access  
In this paper, we propose the multiscale region-based convolutional neural network (MR-CNN).  ...  Small traffic sign recognition is a challenging problem in computer vision, and its accuracy is important to the safety of intelligent transportation systems (ITS).  ...  FIGURE 1 . 1 The architecture of our proposed multi-scale region-based convolutional neural network (MR-CNN).  ... 
doi:10.1109/access.2019.2913882 fatcat:xi6d6uog2ngnpijmzqe52bm4le

Dense-RefineDet for Traffic Sign Detection and Classification

Chang Sun, Yibo Ai, Sheng Wang, Weidong Zhang
2020 Sensors  
Additionally, we presented an anchor-design method to provide suitable anchors for detecting small traffic signs.  ...  Experiments using the Tsinghua-Tencent 100K dataset demonstrated that Dense-RefineDet achieved competitive accuracy at high-speed detection (0.13 s/frame) of small-, medium-, and large-scale traffic signs  ...  [5] proposed MR-CNN, with contextual regions selected based on concatenated features in the MR-CNN to provide additional information for small objects. Noh et al.  ... 
doi:10.3390/s20226570 pmid:33213025 pmcid:PMC7698555 fatcat:g5qmplbq5zh6pgufrdrgboemcq

A lightweight model for multi-traffic object detection based on deep learning under complex traffic conditions

Guoqiang Chen, Yanan Cheng
2021 Journal of Applied Science and Engineering  
The paper presents how to achieve precise results in multi-traffic object detection while minimizing the model size. A deep learning network YOLOv5s-Ghost-SE-DW is proposed based on the YOLOv5s.  ...  The proposed network can detect all traffic objects including traffic signs and lights. First, the convolution layer is replaced by the Ghost module to reduce the parameter and model size.  ...  MR-CNN uses a more accurate Region Proposal Network to transform the region before inputting, pays more attention to its context information and completes better screening and recommendation of the region  ... 
doi:10.6180/jase.202206_25(3).0019 doaj:457599980d104da9a852c81d3e6f55d5 fatcat:cr4ozugtdrbdhkvnafbu7ocwuq

Traffic Sign Detection and Recognition Using Novel Center-Point Estimation and Local Features

Lijing Wei, Cheng Xu, Siqi Li, Xiaohan Tu
2020 IEEE Access  
In this paper, we regard traffic sign detection as a region classification problem and propose a two-stage CNN-based approach to solve it.  ...  INDEX TERMS Traffic sign detection, multi-scale, center-point estimation, local features.  ...  MR-CNN [16] uses the multi-scale features and the surrounding context information of the candidate objects to detect traffic signs, but it ignores the important local features.  ... 
doi:10.1109/access.2020.2991195 fatcat:bgjdr4v6qzhwpla24wgkgermzq

Focus First: Coarse-to-Fine Traffic Sign Detection With Stepwise Learning

Liman Liu, Yuntao Wang, Kunqian Li, Jie Li
2020 IEEE Access  
[33] proposed a multi-scale region-based convolutional neural network (MR-CNN) for small traffic sign detection.  ...  [7] constructed a multi-column cascading convolutional neural network, which further refreshed the recognition effect of traffic signs. Jin et al.  ... 
doi:10.1109/access.2020.3024583 fatcat:uxasq5ufhvbjrmh7ijfb2aj6zy

FAMN: Feature Aggregation Multipath Network for Small Traffic Sign Detection

Zhonghong Ou, Fenrui Xiao, Baiqiao Xiong, Shenda SHI, Meina Song
2019 IEEE Access  
For example, different speed-limit traffic signs have differences solely from the speed numbers.  ...  INDEX TERMS Traffic sign detection, small object detection, fine-grained classification.  ...  It outperforms MR-CNN [17] by 3.2%, 2.2%, and 4.3%, for small, medium, and large size groups, respectively, and by 3.2% on average.  ... 
doi:10.1109/access.2019.2959015 fatcat:amlp5nawkzaevevnjged66s2hi

A Survey on Deep Domain Adaptation and Tiny Object Detection Challenges, Techniques and Datasets [article]

Muhammed Muzammul, Xi Li
2021 arXiv   pre-print
., convolutions and convolutional neural networks (CNN), pooling operations with trending types.  ...  In part 2) we mainly focused on tiny object detection techniques (multi-scale feature learning, Data augmentation, Training strategy (TS), Context-based detection, GAN-based detection).  ...  multi-scale deep convolutional neural network for fast object detection MSCNN [152] A MultiPath network for object detection MPNet [153] Gated bi-directional CNN for object detection GBDNet [154] Contextual  ... 
arXiv:2107.07927v1 fatcat:pgwxu5tnvzhj7ln3ccndmpilsi

Exploring deep learning-based architecture, strategies, applications and current trends in generic object detection: A comprehensive review

Lubna Aziz, Sah bin Haji Salam, Sara Ayub
2020 IEEE Access  
(e.g., StuffNet, HyperNet), modified classification network (e.g., NOC), multi-region and multi-scale feature extraction (e.g., MR-CNN).  ...  Real-time accurate traffic sign recognition helps drive by acquiring temporal and spatial information of the potential sign.  ... 
doi:10.1109/access.2020.3021508 fatcat:guri46oiejhfzeitxuuprpmjka

A Deep Spatial Contextual Long-term Recurrent Convolutional Network for Saliency Detection [article]

Nian Liu, Junwei Han
2016 arXiv   pre-print
In this paper, we propose a novel computational saliency model, i.e., deep spatial contextual long-term recurrent convolutional network (DSCLRCN) to predict where people looks in natural scenes.  ...  Moreover, we also integrate scene context modulation in DSLSTM for saliency inference, leading to a novel deep spatial contextual LSTM (DSCLSTM) model.  ...  ., traffic signs in street views as shown in (c)) or some exceptional objects (e.g., a bed in a forest as shown in (d)).  ... 
arXiv:1610.01708v1 fatcat:lokren3j3nb67axandywdt4b54

Visual saliency prediction based on deep learning

Bashir Ghariba
In Chapter 4, a novel deep learning model based on a Fully Convolutional Network (FCN) architecture is proposed.  ...  The outcomes that have been achieved by neural network methods in a variety of tasks have highlighted their ability to predict visual saliency.  ...  Therefore, based on the advantages of CNNs, we can use small and high receptive fields in down-sampling (e.g., multi-convolution layers, such as in the VGG-16 network) to create feature maps.  ... 
doi:10.48336/bzj8-n176 fatcat:gbladtscczhhnezlp4lxvnwf2u