StrawberryScan(硕果鉴)

Jun 16, 2026 · 1 min read

基于 AIPC 与多模态大模型的端侧 AI 草莓品控工具。结合 YOLO11 目标检测/分割、RAG 知识库检索与多模态大模型,在搭载骁龙 X Elite 的 AIPC 端侧实现草莓成熟度与病虫害的精准识别,可提升优果率 20% 以上。

✨ 核心特性

  • 🍓 成熟度识别:准确率 >95%,辅助采摘决策
  • 🐛 病虫害预警:准确率 >90%,提前发现病害风险
  • 📄 AI 品鉴报告:可视化 + 文化感,提供数字品质证明
  • 端侧部署:运行于 AIPC,支持 NPU 加速,零延迟、高隐私
  • 📚 RAG 增强检索:基于草莓知识库的智能问答与多策略检索

🏗️ 技术栈

类别 技术选型
前端界面 Qt6 (PySide6) + QML
AI 模型 YOLO11 + Qwen2.5
深度学习框架 PyTorch + Ultralytics
RAG 系统 BGE 嵌入模型 + ChromaDB
部署环境 骁龙 X Elite AIPC,支持 NPU 推理加速

🚀 快速开始

# 克隆并安装
git clone https://github.com/SirupLi-C/strawberry-scan.git
cd strawberry-scan
pip install -e .

# 启动应用
python -m strawberry_scan

更多信息请参考项目仓库 README

Justin Hoew
Authors
Machine Learning Specialist
Hoew is a Machine Learning Specialist with deep expertise in reinforcement learning, modern machine learning techniques, and data-driven decision systems. He has extensive hands-on experience developing autonomous driving prediction algorithms, deploying production-grade models, and optimizing computational performance. Passionate about designing robust, scalable AI systems that deliver real-world impact.