{"slug": "rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded", "title": "RAMS: Resource-Adaptive and Detection-Conditioned Model Switching for Embedded Edge Perception", "summary": "Researchers introduced RAMS, a runtime controller that dynamically switches between YOLOv8 model tiers on embedded devices to balance latency and detection quality under resource pressure. On Jetson Orin TensorRT under heavy load, the safety2 policy achieved 3.41 ms mean latency, 5.6x faster than fixed-MEDIUM inference, while retaining 74% of its proxy accuracy. Detection-conditioned switching improved the VRU-Weighted Accuracy Score by up to 47.3% over threshold-only policies.", "body_md": "arXiv:2606.14716v1 Announce Type: new\nAbstract: 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.", "url": "https://wpnews.pro/news/rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded", "canonical_source": "https://arxiv.org/abs/2606.14716", "published_at": "2026-06-16 04:00:00+00:00", "updated_at": "2026-06-16 04:18:09.842921+00:00", "lang": "en", "topics": ["computer-vision"], "entities": ["RAMS", "YOLOv8", "Raspberry Pi 5", "Jetson Orin", "ONNX", "TensorRT", "KITTI"], "alternates": {"html": "https://wpnews.pro/news/rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded", "markdown": "https://wpnews.pro/news/rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded.md", "text": "https://wpnews.pro/news/rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded.txt", "jsonld": "https://wpnews.pro/news/rams-resource-adaptive-and-detection-conditioned-model-switching-for-embedded.jsonld"}}