Unverified75% confidenceFactTime unknown
YOLOv8-nano模型在中端NPU上可实现30FPS以上的推理帧率
1
Sources
75%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified轻量化模型通常采用YOLOv8-nano、MobileNet等经剪枝量化蒸馏处理的网络，在低分辨率下实现每秒30帧以上实时推理77% similarUnverifiedYOLOv8 模型分为 nano 到 x 多个规格，YOLOv8s 在边缘设备上可实现实时推理77% similarUnverifiedMoS-TTS-Nano可能采用了非自回归架构来平衡质量与推理速度65% similarUnverifiedYOLOv8 模型家族从 YOLOv8n（约 3.2M 参数）到 YOLOv8x（约 68M 参数）覆盖从树莓派到高性能 GPU 的硬件谱系64% similarUnverifiedAI边缘推理场景选带NPU的板子：RK3588(6TOPS)适合多路视频分析，Jetson系列(Orin Nano 40TOPS起)适合CUDA生态和PyTorch/TensorRT模型直接部署63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/25905API
curl https://kongchang.com/api/v1/knowledge/claims/25905MCP
get_claim(id=25905)