嵌入式设备上动态切换YOLO模型，兼顾速度与检测安全。
RAMS: Resource-Adaptive and Detection-Conditioned Model Switching for Embedded Edge Perception
- 根据设备负载实时切换YOLOv8不同大小的模型，无需重加载。
- 在高负载下平均延迟仅3.41毫秒，保留74%的检测精度。
- 针对行人等易漏检目标优化切换策略，提升关键场景召回率。
嵌入式硬件上的边缘目标检测需在推理延迟与检测质量间权衡资源压力变化。本文提出RAMS，一种轻量级运行时控制器，可监测设备压力，基于空闲行为校准切换阈值，并在三个驻留的YOLOv8层级（NANO/SMALL/MEDIUM，对应320/416/640像素）间动态选择，无模型重载延迟。RAMS定义了五种切换策略，包括两种检测条件型变体，防止在检测到脆弱道路使用者（VRU）后激进降级。我们引入VRU加权准确率得分（SWAS），一个无需真实标注即可离线比较策略的标量指标，并提出一个基于理想基准的变体，以分离检测器循环性与真实层级保留收益。在Raspberry Pi 5、x86笔记本和Jetson Orin ONNX/TensorRT部署中，同一控制器方程覆盖37倍延迟范围。在Jetson Orin TensorRT高负载下，safety2策略实现3.41毫秒平均延迟，比固定使用MEDIUM快5.6倍，同时通过近NANO操作保留74%的代理准确率，且在检测到VRU时选择性锁定SMALL和MEDIUM层级。检测条件切换相比仅阈值策略，在理想评分下提升SWAS 25.4%，在检测器推导评分下提升47.3%。实际KITTI评估显示各层级的VRU召回率为24.2%、41.2%、59.0%，表明反应式覆盖受限于基础检测器召回能力。
原文摘要 · Abstract (English)
Edge object detection on embedded hardware requires balancing inference latency and detection quality under changing resource pressure. We present RAMS, a lightweight runtime controller that monitors device pressure, calibrates switching thresholds from idle behavior, and dynamically selects among three resident YOLOv8 tiers (NANO/SMALL/MEDIUM at 320/416/640 px) without model-reload latency. RAMS defines five switching policies, including two detection-conditioned variants that prevent aggressive downgrades after recent vulnerable-road-user (VRU) detections. We further introduce the VRU-Weighted Accuracy Score (SWAS), a scalar metric for offline policy comparison without ground-truth annotations, together with an oracle-bounded variant that separates detector circularity from genuine tier-retention benefit. Across Raspberry Pi 5, x86 laptops, and Jetson Orin ONNX/TensorRT deployments, the same controller equations operate over a 37x latency range. On Jetson Orin TensorRT under heavy load, the safety2 policy achieves 3.41 ms mean latency, 5.6x faster than fixed-MEDIUM inference, while retaining 74% of its proxy accuracy through near-NANO operation with selective SMALL and MEDIUM locks during VRU-positive windows. Detection-conditioned switching improves SWAS by 25.4% under oracle scoring and 47.3% under detector-derived scoring relative to threshold-only policies under heavy load. Live KITTI evaluation reports per-tier VRU recall of 24.2%, 41.2%, and 59.0%, showing that reactive overrides are fundamentally limited by baseline detector recall.
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。