Scholarlink

<p>License plate type classification, which categorizes vehicles by their plate color (e.g., green for new-energy vehicles), is an important component of intelligent transportation systems for differentiated management and analysis. To address the low efficiency of traditional methods in complex environments, we propose a deep learning license plate type classification model based on an improved AlexNet and dual attention mechanisms (ADAM). First, the classical AlexNet is optimized to promote feature extraction efficiency. Second, channel and spatial dual attention mechanisms are incorporated to enhance feature learning ability. Finally, an efficient channel attention mechanism is added to improve classification accuracy. The experimental results demonstrate that the proposed method achieves a recognition accuracy of 99.25%, which is superior to other excellent deep learning networks such as VGG16, EfficientNetV2, and ZFNet. In addition to having some reference value for improving the utility and universality of intelligent transportation systems, it has the potential to expand multi-language and multi-scenario license plate type classification employment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ADAM: A dual attention mechanism deep learning neural network for license plate type classification

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Qiuying Han,
  • Lirani He,
  • Shaohui Zhang,
  • Peng Wang,
  • Laixiang Xu

摘要

License plate type classification, which categorizes vehicles by their plate color (e.g., green for new-energy vehicles), is an important component of intelligent transportation systems for differentiated management and analysis. To address the low efficiency of traditional methods in complex environments, we propose a deep learning license plate type classification model based on an improved AlexNet and dual attention mechanisms (ADAM). First, the classical AlexNet is optimized to promote feature extraction efficiency. Second, channel and spatial dual attention mechanisms are incorporated to enhance feature learning ability. Finally, an efficient channel attention mechanism is added to improve classification accuracy. The experimental results demonstrate that the proposed method achieves a recognition accuracy of 99.25%, which is superior to other excellent deep learning networks such as VGG16, EfficientNetV2, and ZFNet. In addition to having some reference value for improving the utility and universality of intelligent transportation systems, it has the potential to expand multi-language and multi-scenario license plate type classification employment.