Table 3.
Summary of Deep Learning Approaches to Traffic Sign Recognition.
| Author | Algorithm | Dataset | Accuracy (%) |
|---|---|---|---|
| Siniosoglou (2021) [10] | Deep Autoencoder | CATERED | 99.19 |
| Li and Wang (2018) [11] | MobileNet | GTSRB | 99.66 |
| Zhu and Yan (2022) [12] | YOLOv5 | Self-collected | 97.70 |
| Shustanov and Yakimov (2017) [13] | CNN | GTSRB | 99.94 |
| Alghmgham et al., (2019) [14] | CNN | Self-collected (Arabic) | 100 |
| Li et al., (2019) [15] | CNN | GTSRB | 97.4 |
| BTSD | 98.1 | ||
| Yazdan and Varshosaz (2021) [16] | Normalized cross-correlation (NCC) | BTSD | 93.10 |
| Bangquan and Xiong (2019) [17] | LeNet | GTSRB | 98.6 |
| VGG16 | 96.7 | ||
| Zaibi et al., (2021) [18] | LeNet-5 | GTSRB | 99.84 |
| BTSD | 98.37 | ||
| Sreya (2021) [19] | CNN (LeNet) | GTRSB | 90.07 |
| Abudhagir and Ashok (2022) [20] | CNN (LeNet) | GTRSB | 98.50 |
| Mehta et al., (2019) [21] | CNN | BTSD | 97.06 |
| Zhang et al., (2020) [22] | CNN | GTSRB | 99.38 |
| BTSC | 98.89 | ||
| Sokipriala and Orike (2021) [23] | ResNet50 | GTSRB | 95.4 |
| VGG16 | 95.5 | ||
| CNN | 96.0 | ||
| Vincent et al., (2020) [24] | CNN | GTSRB | 98.44 |
| Madan et al., (2019) [25] | Basic CNN | GTSRB | 98.07 |
| Branching CNN | 98.48 | ||
| Serna and Ruichek (2018) [26] | CNN | GTSRB | 99.37 |
| ETSD | 98.99 | ||
| Mishra and Goyal (2022) [27] | CNN | GTSRB | 99.76 |
| BTSC | 99.79 | ||
| TSRD + GTSRB | 98.37 | ||
| Chen e t al. (2017) [28] | MCNN | GTSRB | 97.96 |
| MCNN | 98.26 | ||
| Zheng and Jiang (2022) [29] | DenseNet | GTSRB | 98.82 |
| CCTSDB | 99.42 | ||
| ShuffleNet | ICTS | 99.11 | |
| RealFormer | GTSRB | 86.03 | |
| TNT | CCTSDB | 95.05 | |
| Haque et al., (2021) [30] | DeepThin | GTSRB | 99.72 |
| BTSC | 99.29 | ||
| Siniosoglou (2021) [10] | Deep Autoencoder | CATERED | 99.19 |
| Usha et al., (2021) [31] | CNN | GTSRB | 97.80 |
| Fang et al., (2022) [32] | MicronNet-BF | GTSRB | 99.38 |
| Sarku et al., (2021) [33] | ResNet18 | Self-collected | 60 |
| ResNet50 | 93 | ||
| ResNet152 | 33 | ||
| Cao et al., (2019) [34] | LeNet-5 | GTSRB | 99.75 |
| Fu and Wang (2021) [35] | MSCN + MCDNN | TSRD (train), GTSRB (test) | 90.13 |
| Sichkar and Kolyubin (2019) [36] | CNN | GTSRB | 86.80 |
| Agarwal et al., (2022) [37] | CNN | GTSRB | 99.66 |
| Youssouf (2022) [38] | CNN | GTSRB | 99.20 |
| Gökberk et al. (2022) [39] | AlexNet | GTSRB | 97.45 |
| DarkNet-53 | 94.69 | ||
| EfficientNet-b0 | 98.64 | ||
| Kuros and Kryjak (2022) [40] | DNN | GTSRB | 99.86 |
| QNN | 94.40 | ||
| Pradana et al., (2022) [41] | CNN | Indonesian Traffic Signs | 93.00 |
| Bhatt et al., (2022) [42] | CNN | GTSRB | 99.85 |
| Self-collected (Indian), | 91.08 | ||
| GTSRB+Indian | 95.45 | ||
| Mamatkulovich (2022) [43] | CNN | GTSRB | 99.93 |