| 王振宇,赵科东,孙永荣,杨杰,陈旗.基于CNN-LRP层级适应性冻结的模型优化方法[J].电子测量与仪器学报,2026,40(5):286-295 |
| 基于CNN-LRP层级适应性冻结的模型优化方法 |
| LATS: Model optimisation method based on CNN-LRPlayer-wise adaptability freezing |
| |
| DOI： |
| 中文关键词: 迁移学习 小样本学习 域适应 信号灯状态监测 模型微调 模型优化 |
| 英文关键词:transfer learning model fine-tuning small sample learning domain adaptation traffic light status monitoring model optimisation |
| 基金项目: |
|
|
| Author | Institution |
| Wang Zhenyu | Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| Zhao Kedong | Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| Sun Yongrong | Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| Yang Jie | Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| Chen Qi | Navigation Research Center, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
|
| 摘要点击次数: 162 |
| 全文下载次数: 76 |
| 中文摘要: |
| 针对外观多样且样本稀缺的信号灯检测场景，迁移学习已成为主流解决方案。然而，传统全参数微调和浅层冻结未量化网络层对任务具体贡献，训练更新范围过广、参数冗余，在小样本条件下易过拟合从而限制模型性能。为此提出一种基于卷积神经网络 分层相关性传播（CNN-LRP）层级适应性冻结的模型优化方法。基于正向相关比、空间聚焦度与层深惩罚因子，将CNN-LRP的像素级归因拓展至卷积层级，构建层级贡献度评估体系，实现解释性分析与冻结策略的融合。针对传统策略任务适应性不足的问题，设计了多指标融合的卷积层适应性评分方法，自动生成冻结策略，通过冻结低适应性层，实现任务驱动的迁移，从而优化模型。最后，利用不同光照与距离的场景数据进行验证测试。结果表明，相比于全微调与其他冻结策略，该方法生成的模型在测试准确率上提升达3.1%，参数调优量下降34.49%，在精度、效率、泛化性、鲁棒性与卷积网络普适性等方面具有更好的效果。 |
| 英文摘要: |
| In multi-scenario perception and intelligent control systems, the accurate and efficient recognition of traffic light states is an important support for ensuring stable system operation and the collaborative operation of intelligent devices. In response to the diverse appearances and scarce samples in traffic light detection scenarios, transfer learning has become a mainstream solution. However, traditional full parameter fine-tuning and shallow freezing do not quantify the specific contributions of network layers to the task, leading to overly broad training updates, parameter redundancy, and a tendency to overfit under small sample conditions, which limits model performance. To address this, this paper proposes a layer-adaptive guided freezing strategy generation method based on CNN-LRP. By utilising the forward correlation ratio, spatial focus degree, and layer depth penalty factor, the pixel-level correlation of CNN-LRP is expanded to the convolutional layer level, constructing a layer contribution evaluation system that integrates interpretative analysis with the freezing strategy. To tackle the issue of insufficient task adaptability in traditional strategies, this paper designs a multi-metric fusion convolutional layer adaptability scoring method that automatically generates freezing strategies, and realises task-driven migration by freezing the low-adaptability layer, so as to optimise the model. Finally, verification tests were conducted using scene data with different lighting and distances. The results show that compared with full fine-tuning and other freezing strategies, the model generated by the method proposed in this paper has an improvement of 3.1% in test accuracy, a reduction of 34.49% in parameter tuning volume, and better performance in terms of accuracy, efficiency, generalization, robustness, and the universality of convolutional networks. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
