用知识蒸馏让小模型在边缘设备上精准检测行人，误报率降44%
Edge AI for Automotive Vulnerable Road User Safety: Deployable Detection via Knowledge Distillation
- 用大模型教小模型，实现3.9倍压缩
- 小模型在量化后误报率降低44%，精度反超大模型
- 适合做车载安全系统的边缘部署
在边缘硬件上部署高精度的弱势道路使用者（VRU）检测面临模型容量与计算约束的平衡难题。大模型虽准确但经INT8量化后性能崩溃，小模型则牺牲检测能力。本文提出一种知识蒸馏（KD）框架，训练一个参数为11.2M的YOLOv8-S学生模型模仿参数为43.7M的YOLOv8-L教师模型，在保持量化鲁棒性的同时实现3.9倍压缩。在包含7万张训练图像的BDD100K数据集上评估，教师模型在INT8量化下mAP下降23%，而KD学生模型仅下降5.6%。分析表明，知识蒸馏传递的是精度校准而非原始检测能力：在相同召回率下，KD学生模型的精度达0.748，优于直接训练的0.653（提升14.5%），使误报减少44%；且其在INT8下的精度（0.748）超过教师模型在FP32下的精度（0.718），模型规模仅为后者的1/3.9。这些结果确立了知识蒸馏在边缘部署高精度、安全关键的VRU检测中的必要性。
原文摘要 · Abstract (English)
Deploying accurate object detection for Vulnerable Road User (VRU) safety on edge hardware requires balancing model capacity against computational constraints. Large models achieve high accuracy but fail under INT8 quantization required for edge deployment, while small models sacrifice detection performance. This paper presents a knowledge distillation (KD) framework that trains a compact YOLOv8-S student (11.2M parameters) to mimic a YOLOv8-L teacher (43.7M parameters), achieving 3.9x compression while preserving quantization robustness. We evaluate on full-scale BDD100K (70K training images) with Post-Training Quantization to INT8. The teacher suffers catastrophic degradation under INT8 (-23% mAP), while the KD student retains accuracy (-5.6% mAP). Analysis reveals that KD transfers precision calibration rather than raw detection capacity: the KD student achieves 0.748 precision versus 0.653 for direct training at INT8, a 14.5% gain at equivalent recall, reducing false alarms by 44% versus the collapsed teacher. At INT8, the KD student exceeds the teacher's FP32 precision (0.748 vs. 0.718) in a model 3.9x smaller. These findings establish knowledge distillation as a requirement for deploying accurate, safety-critical VRU detection on edge hardware.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。