[论文解读] Modified Supervised Contrastive Learning for Detecting Anomalous Driving Behaviours.

Shehroz S. Khan, Ziting Shen+3 more|arXiv (Cornell University)|Sep 9, 2021

Anomaly Detection Techniques and Applications参考文献 25被引用 4

一句话总结

本文提出了一种改进的监督对比学习框架，通过在推理过程中保留投影头并重新加权对比损失中的负样本对，增强了对已见和未见异常驾驶行为的视觉表征学习能力。该方法在783分钟的驾驶数据集上，利用来自俯视和前视摄像头的融合深度与红外模态，实现了SOTA性能，AUC ROC峰值达0.9738，AUC PR达0.9772。

ABSTRACT

Detecting distracted driving behaviours is important to reduce millions of deaths and injuries occurring worldwide. Distracted or anomalous driving behaviours are deviations from the 'normal' driving that need to be identified correctly to alert the driver. However, these driving behaviours do not comprise of one specific type of driving style and their distribution can be different during training and testing phases of a classifier. We formulate this problem as a supervised contrastive learning approach to learn a visual representation to detect normal, and seen and unseen anomalous driving behaviours. We made a change to the standard contrastive loss function to adjust the similarity of negative pairs to aid the optimization. Normally, the (self) supervised contrastive framework contains an encoder followed by a projection head, which is omitted during testing phase as the encoding layers are considered to contain general visual representative information. However, we assert that for supervised contrastive learning task, including projection head will be beneficial. We showed our results on a Driver Anomaly Detection dataset that contains 783 minutes of video recordings of normal and anomalous driving behaviours of 31 drivers from various from top and front cameras (both depth and infrared). We also performed an extra step of fine tuning the labels in this dataset. Out of 9 video modalities combinations, our modified contrastive approach improved the ROC AUC on 7 in comparison to the baseline models (from 3.12% to 8.91% for different modalities); the remaining two models also had manual labelling. We performed statistical tests that showed evidence that our modifications perform better than the baseline contrastive models. Finally, the results showed that the fusion of depth and infrared modalities from top and front view achieved the best AUC ROC of 0.9738 and AUC PR of 0.9772.

研究动机与目标

• 解决检测多样且未见的异常驾驶行为的挑战，这些行为偏离了正常驾驶模式。
• 克服训练与测试数据中异常行为分布偏移的问题，这限制了传统监督模型的性能。
• 通过在监督设置下适应对比学习并引入动态负样本对加权机制，改进视觉表征学习以提升异常检测性能。
• 证明在监督对比学习中推理时保留投影头具有优势，与标准实践相反。
• 在多种视频模态组合下实现稳健性能，包括对未见异常行为的检测。

提出的方法

• 提出一种改进的对比损失函数，通过调整负样本对的相似度来提升优化稳定性与表征学习能力。
• 在推理过程中保留投影头，认为其有助于提升监督对比学习中异常检测的特征判别能力。
• 对Driver Anomaly Detection数据集的标签进行微调，以提升标注一致性与模型泛化能力。
• 使用编码器与投影头在来自前视与俯视摄像头的多模态输入（深度、红外、RGB）上训练模型。
• 对来自两个摄像头视角的深度与红外模态进行早期融合，以增强判别性特征学习。
• 利用对比学习学习能将正常行为与已见及未见异常行为区分开的表征。

实验结果

研究问题

• RQ1 与标准对比学习相比，改进的监督对比学习方法是否能提升对已见和未见异常驾驶行为的检测性能？
• RQ2 在推理过程中保留投影头是否能提升驾驶行为识别中的异常检测性能？
• RQ3 来自俯视与前视摄像头的深度与红外模态融合如何影响检测性能？
• RQ4 在对比损失中重新加权负样本对在多大程度上改善了模型的优化与泛化能力？
• RQ5 对数据集标签的微调是否能在异常检测任务中带来可测量的性能提升？

主要发现

• 与基线模型相比，改进的对比方法在不同视频模态组合下，AUC ROC提升了3.12%至8.91%。
• 来自俯视与前视视角的深度与红外模态融合实现了最高性能，AUC ROC达0.9738，AUC PR达0.9772。
• 统计检验证实，所提改进方法在多种配置下显著优于基线对比模型。
• 该方法在9种视频模态组合中的7种上实现了性能提升，其余两种模型也因人工标签优化而受益。
• 结果表明，推理时包含投影头具有显著优势，挑战了传统上训练后丢弃投影头的做法。
• 该模型在未见异常行为上表现出强大的泛化能力，表明其在真实部署场景中具备鲁棒性。