一个基于训练有素的YOLOv8模型用于检测作物叶片病害的全栈网页应用。

特色

• 上传叶片图像，并从训练好的模型中获取典型疾病标签。
• 前端仪表盘，用于扫描历史、洞察和结果。
• 后端 API，用于上传、推理和持久化。

技术栈

• 前端：React 18 + TypeScript、Vite、Tailwind CSS、Radix UI（shadcn 风格组件）。
• 后端：Node.js + TypeScript、Express、tsx，用于直接运行TS。
• 数据库：通过 Drizzle ORM 和 drizzle-kit 迁移的 PostgreSQL。
• 机器学习：Python（Ultralytics YOLOv8、PyTorch、ONNX 运行时可选）。
• 开发工具：Vite、TypeScript、tsx、tailwindcss、cross-env.

快速入门（开发） 前提条件

• 节点18+和npm
• Python 3.10+（机器学习工具）
• PostgreSQL（或兼容的DATABASE_URL）

1. 克隆仓库

2. 安装节点依赖

3. 配置环境

• 在项目根创建一个.env，至少满足：

4. （可选）为机器学习准备 Python 环境

5. 运行数据库迁移

6. 启动开发中的应用

dev 脚本运行 TypeScript 服务器（参见 server/index.ts），该服务器在开发过程中也为客户端提供服务。

量产版本

ML-POWERED TOMATO CROP DISEASE IDENTIFICATION AND MONITORING

A full-stack web application for crop leaf disease detection using a trained YOLOv8 model.

Features

• Upload leaf images and get a canonical disease label from the trained model.
• Frontend dashboard for scan history, insights and results.
• Backend API for uploads, inference and persistence.

Tech stack

• Frontend: React 18 + TypeScript, Vite, Tailwind CSS, Radix UI (shadcn-style components).
• Backend: Node.js + TypeScript, Express, tsx for running TS directly.
• Database: PostgreSQL via Drizzle ORM and drizzle-kit migrations.
• Machine learning: Python (Ultralytics YOLOv8, PyTorch, ONNX Runtime optional).
• Dev tools: Vite, TypeScript, tsx, tailwindcss, cross-env.

Quickstart (development) Prerequisites

• Node 18+ and npm
• Python 3.10+ (for ML tooling)
• PostgreSQL (or a compatible DATABASE_URL)

1. Clone the repo

git clone  && cd MP

2. Install Node dependencies

npm install

3. Configure environment

• Create a .env at the project root with at least:

DATABASE_URL=postgres://user:pass@host:port/dbname
SESSION_SECRET=your_session_secret
# Optional: OPENAI_API_KEY=...

4. (Optional) Prepare Python environment for ML

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r ml/requirements.txt

5. Run database migrations

npm run db:push

6. Start the app in development

npm run dev

The dev script runs the TypeScript server (see server/index.ts) which also serves the client during development.

Production build

npm run build
npm start

Machine learning / inference

• The trained model expected by the server is at ml/models/best.pt (or the provided yolov8n.pt for experiments).
• To run a quick inference locally:

python ml/predict.py path/to/image.jpg

• To evaluate the dataset splits:

python ml/predict.py --eval test [limit]

Troubleshooting

• If you see “DATABASE_URL, ensure the database is provisioned” the .env or PostgreSQL connection is not configured correctly.
• Model errors will indicate missing ml/models/best.pt — ensure the file exists before running inference.

Where to look in the repo

• Server code: server/index.ts
• Frontend: client/src
• ML: ml (training, inference and requirements)