fix: normalize repository name to lowercase in deployment workflow
This commit is contained in:
Asep Haryana Saputra
Submodule .claude/worktrees/agent-a535691b7def5bafe deleted from 80df164a2b
@@ -8,7 +8,7 @@ on:
|
||||
|
||||
env:
|
||||
REGISTRY: ghcr.io
|
||||
IMAGE_PREFIX: ghcr.io/${{ github.repository }}
|
||||
IMAGE_PREFIX: ghcr.io/${{ toLower(github.repository) }}
|
||||
|
||||
jobs:
|
||||
build:
|
||||
@@ -101,7 +101,7 @@ jobs:
|
||||
git pull --ff-only origin main
|
||||
|
||||
cat > .env << 'ENVEOF'
|
||||
GITHUB_REPOSITORY=${{ github.repository }}
|
||||
GITHUB_REPOSITORY=${{ toLower(github.repository) }}
|
||||
DATABASE_URL=${{ secrets.DATABASE_URL }}
|
||||
SESSION_SECRET=${{ secrets.SESSION_SECRET }}
|
||||
WEB_APP_URL=https://zeavisedu.asepharyana.tech
|
||||
|
||||
@@ -0,0 +1,382 @@
|
||||
# ZeaVis Edu
|
||||
|
||||
ZeaVis Edu adalah aplikasi fullstack untuk klasifikasi penyakit daun jagung menggunakan machine learning. Sistem ini terdiri dari tiga layanan utama: web frontend, API backend, dan layanan ML, yang dapat dijalankan secara lokal atau di-deploy dengan Docker.
|
||||
|
||||
## Fitur
|
||||
|
||||
- **Klasifikasi Penyakit Daun Jagung**: Mengidentifikasi 4 jenis penyakit daun jagung
|
||||
- **Antarmuka Web Modern**: Dibangun dengan React, Vite, dan TypeScript
|
||||
- **API RESTful**: Backend Elysia dengan Drizzle ORM untuk PostgreSQL
|
||||
- **Layanan ML Terpisah**: FastAPI dengan model TensorFlow EfficientNetV2B0
|
||||
- **Docker Deployment**: Containerized dengan docker-compose dan Traefik reverse proxy
|
||||
- **Workflow ML Lengkap**: Dari preprocessing data hingga export model produksi
|
||||
|
||||
## Kelas Penyakit yang Dideteksi
|
||||
|
||||
Model machine learning dapat mengklasifikasikan 4 kondisi daun jagung:
|
||||
|
||||
1. **Bercak Daun** — Gray Leaf Spot
|
||||
2. **Hawar Daun** — Northern/Southern Leaf Blight
|
||||
3. **Karat Daun** — Common Rust
|
||||
4. **Daun Sehat** — healthy corn leaf
|
||||
|
||||
## Struktur Proyek
|
||||
|
||||
```
|
||||
ZeaVis-Edu/
|
||||
├── apps/
|
||||
│ ├── web/ # Frontend React + Vite + TypeScript
|
||||
│ ├── api/ # Backend Elysia + Drizzle + PostgreSQL
|
||||
│ └── ml-service/ # FastAPI + TensorFlow ML service
|
||||
├── packages/
|
||||
│ └── shared/ # Shared TypeScript types dan utilities
|
||||
├── Machine_Learning/ # Pipeline ML: preprocessing, training, export
|
||||
├── docker-compose.yml
|
||||
└── package.json # Root workspace dengan Moon tasks
|
||||
```
|
||||
|
||||
## Tech Stack
|
||||
|
||||
### Frontend (apps/web)
|
||||
- **React 18** dengan TypeScript
|
||||
- **Vite** untuk build tool
|
||||
- **React Router** untuk routing
|
||||
- **TanStack Query** untuk data fetching
|
||||
- **Zustand** untuk state management
|
||||
- **Tailwind CSS** untuk styling
|
||||
- **shadcn/ui** untuk komponen UI
|
||||
|
||||
### Backend (apps/api)
|
||||
- **Bun** runtime
|
||||
- **Elysia** framework web
|
||||
- **Drizzle ORM** dengan PostgreSQL
|
||||
- **TypeScript**
|
||||
|
||||
### ML Service (apps/ml-service)
|
||||
- **FastAPI** dengan Python
|
||||
- **TensorFlow** untuk inferensi model
|
||||
- **EfficientNetV2B0** arsitektur model
|
||||
- **Pydantic** untuk validasi data
|
||||
|
||||
### ML Pipeline (Machine_Learning/)
|
||||
- **Python** dengan TensorFlow/Keras
|
||||
- **EfficientNetV2B0** untuk training
|
||||
- **Google Colab** untuk training dengan GPU
|
||||
- **TensorFlow SavedModel, TFLite, TensorFlow.js** untuk export
|
||||
|
||||
### Infrastruktur
|
||||
- **Docker** dan **docker-compose** untuk containerization
|
||||
- **Traefik** sebagai reverse proxy
|
||||
- **PostgreSQL** database
|
||||
- **Moon** sebagai task runner untuk monorepo
|
||||
|
||||
## Prasyarat
|
||||
|
||||
- **Node.js 18+** atau **Bun** (direkomendasikan)
|
||||
- **Python 3.9+** dengan pip
|
||||
- **Docker** dan **docker-compose** (untuk deployment)
|
||||
- **Git**
|
||||
|
||||
## Instalasi
|
||||
|
||||
### 1. Clone Repository
|
||||
```bash
|
||||
git clone https://github.com/mytheclipse/ZeaVis-Edu.git
|
||||
cd ZeaVis-Edu
|
||||
```
|
||||
|
||||
### 2. Install Dependencies
|
||||
```bash
|
||||
bun install
|
||||
```
|
||||
|
||||
### 3. Setup Environment Variables
|
||||
Salin file `.env.example` ke `.env` dan sesuaikan nilai-nilainya:
|
||||
```bash
|
||||
cp .env.example .env
|
||||
```
|
||||
|
||||
## Menjalankan Secara Lokal
|
||||
|
||||
### Perintah Root (Menggunakan Moon)
|
||||
```bash
|
||||
# Development mode (semua layanan)
|
||||
bun run dev
|
||||
|
||||
# Build production
|
||||
bun run build
|
||||
|
||||
# Type checking
|
||||
bun run typecheck
|
||||
```
|
||||
|
||||
### Web App (apps/web)
|
||||
```bash
|
||||
cd apps/web
|
||||
bun run dev
|
||||
```
|
||||
Akses di: http://localhost:5173
|
||||
|
||||
### API (apps/api)
|
||||
```bash
|
||||
cd apps/api
|
||||
bun run start
|
||||
```
|
||||
Akses di: http://localhost:3000
|
||||
|
||||
### ML Service (apps/ml-service)
|
||||
```bash
|
||||
cd apps/ml-service
|
||||
# Install dependencies Python
|
||||
pip install -r requirements.txt
|
||||
# Jalankan service
|
||||
uvicorn main:app --reload --port 8000
|
||||
```
|
||||
Akses di: http://localhost:8000
|
||||
|
||||
## Endpoint ML Service
|
||||
|
||||
### 1. GET /health
|
||||
**Deskripsi**: Health check endpoint
|
||||
**Response**:
|
||||
```json
|
||||
{
|
||||
"status": "healthy",
|
||||
"timestamp": "2024-01-01T00:00:00Z"
|
||||
}
|
||||
```
|
||||
|
||||
### 2. GET /metadata
|
||||
**Deskripsi**: Mendapatkan metadata model
|
||||
**Response**:
|
||||
```json
|
||||
{
|
||||
"model_name": "EfficientNetV2B0",
|
||||
"input_size": 224,
|
||||
"classes": ["Bercak Daun", "Hawar Daun", "Karat Daun", "Daun Sehat"],
|
||||
"version": "1.0.0"
|
||||
}
|
||||
```
|
||||
|
||||
### 3. POST /predict
|
||||
**Deskripsi**: Prediksi gambar daun jagung
|
||||
**Request Body**:
|
||||
```json
|
||||
{
|
||||
"image": "base64_encoded_image_string"
|
||||
}
|
||||
```
|
||||
**Response**:
|
||||
```json
|
||||
{
|
||||
"predictions": [
|
||||
{
|
||||
"class": "Bercak Daun",
|
||||
"confidence": 0.95
|
||||
},
|
||||
{
|
||||
"class": "Hawar Daun",
|
||||
"confidence": 0.03
|
||||
},
|
||||
{
|
||||
"class": "Karat Daun",
|
||||
"confidence": 0.01
|
||||
},
|
||||
{
|
||||
"class": "Daun Sehat",
|
||||
"confidence": 0.01
|
||||
}
|
||||
],
|
||||
"top_prediction": {
|
||||
"class": "Bercak Daun",
|
||||
"confidence": 0.95
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
## Deployment dengan Docker
|
||||
|
||||
### 1. Build dan Jalankan dengan docker-compose
|
||||
```bash
|
||||
docker-compose up -d
|
||||
```
|
||||
|
||||
### 2. Services dalam docker-compose.yml
|
||||
- **web**: Frontend React app (port 80 dalam container)
|
||||
- **api**: Backend API (port 3000 dalam container)
|
||||
- **ml**: ML service (port 8000 dalam container)
|
||||
- **Network**: `app-shared-net` untuk komunikasi antar service
|
||||
|
||||
### 3. Environment Variables untuk Docker
|
||||
Pastikan file `.env` berisi:
|
||||
```env
|
||||
DATABASE_URL=postgresql://user:password@postgres:5432/zeavis
|
||||
API_PORT=3000
|
||||
WEB_APP_URL=https://zeavisedu.asepharyana.tech
|
||||
ML_SERVICE_URL=https://ml.zeavisedu.asepharyana.tech
|
||||
MODEL_PATH=/app/model/best_model.keras
|
||||
MODEL_INPUT_SIZE=224
|
||||
```
|
||||
|
||||
## Workflow Machine Learning
|
||||
|
||||
### 1. Preprocessing Data
|
||||
```bash
|
||||
cd Machine_Learning
|
||||
python preprocessing.py
|
||||
```
|
||||
Membutuhkan file `dataset_1.zip`, `dataset_2.zip`, `dataset_3.zip` di direktori yang sama.
|
||||
|
||||
### 2. Training di Google Colab
|
||||
- Buka `notebook.ipynb` di Google Colab dengan GPU enabled
|
||||
- Upload `dataset.zip` yang dihasilkan dari preprocessing
|
||||
- Jalankan notebook untuk training model EfficientNetV2B0
|
||||
- Model terbaik akan disimpan sebagai `best_model.keras`
|
||||
|
||||
### 3. Export Model untuk Produksi
|
||||
```bash
|
||||
cd Machine_Learning
|
||||
python save_model.py
|
||||
```
|
||||
Menghasilkan:
|
||||
- `model/saved_model/` (TensorFlow SavedModel)
|
||||
- `model/model.tflite` (TFLite format)
|
||||
|
||||
### 4. Convert ke TensorFlow.js
|
||||
```bash
|
||||
export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
|
||||
tensorflowjs_converter \
|
||||
--input_format=tf_saved_model \
|
||||
--output_format=tfjs_graph_model \
|
||||
--signature_name=serving_default \
|
||||
--saved_model_tags=serve \
|
||||
model/saved_model \
|
||||
model/tfjs_model
|
||||
```
|
||||
|
||||
## Artifak yang Dihasilkan
|
||||
|
||||
### File/Direktori yang Dihasilkan (tidak termasuk dalam repo)
|
||||
- `Machine_Learning/dataset_1.zip`, `dataset_2.zip`, `dataset_3.zip` — dataset sumber
|
||||
- `Machine_Learning/dataset/` dan `dataset.zip` — hasil preprocessing
|
||||
- `Machine_Learning/best_model/best_model.keras` — model terlatih dari Colab
|
||||
- `Machine_Learning/model/saved_model/`, `model/model.tflite`, `model/tfjs_model/` — export produksi
|
||||
|
||||
## Environment Variables
|
||||
|
||||
### Umum
|
||||
```env
|
||||
NODE_ENV=development|production
|
||||
```
|
||||
|
||||
### Database
|
||||
```env
|
||||
DATABASE_URL=postgresql://user:password@host:5432/database
|
||||
```
|
||||
|
||||
### API
|
||||
```env
|
||||
API_PORT=3000
|
||||
WEB_APP_URL=http://localhost:5173
|
||||
ML_SERVICE_URL=http://localhost:8000
|
||||
```
|
||||
|
||||
### ML Service
|
||||
```env
|
||||
MODEL_PATH=/path/to/best_model.keras
|
||||
MODEL_INPUT_SIZE=224
|
||||
```
|
||||
|
||||
## Troubleshooting
|
||||
|
||||
### 1. Bun Install Error
|
||||
```bash
|
||||
# Jika bun tidak terinstall
|
||||
curl -fsSL https://bun.sh/install | bash
|
||||
```
|
||||
|
||||
### 2. Python Dependencies Error
|
||||
```bash
|
||||
cd apps/ml-service
|
||||
python -m venv venv
|
||||
source venv/bin/activate
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
|
||||
### 3. Docker Network Error
|
||||
```bash
|
||||
# Buat network jika belum ada
|
||||
docker network create app-shared-net
|
||||
```
|
||||
|
||||
### 4. ML Model Not Found
|
||||
Pastikan file `best_model.keras` ada di:
|
||||
- `Machine_Learning/best_model/best_model.keras` (untuk local)
|
||||
- `/app/model/best_model.keras` (untuk Docker container)
|
||||
|
||||
## Workflow Development
|
||||
|
||||
### 1. Setup Development Environment
|
||||
```bash
|
||||
git clone <repository>
|
||||
cd ZeaVis-Edu
|
||||
bun install
|
||||
cp .env.example .env
|
||||
```
|
||||
|
||||
### 2. Jalankan Layanan Secara Terpisah
|
||||
```bash
|
||||
# Terminal 1: Web app
|
||||
cd apps/web && bun run dev
|
||||
|
||||
# Terminal 2: API
|
||||
cd apps/api && bun run start
|
||||
|
||||
# Terminal 3: ML service
|
||||
cd apps/ml-service && uvicorn main:app --reload --port 8000
|
||||
```
|
||||
|
||||
### 3. Testing
|
||||
```bash
|
||||
# Type checking
|
||||
bun run typecheck
|
||||
|
||||
# Build production
|
||||
bun run build
|
||||
```
|
||||
|
||||
### 4. Docker Testing
|
||||
```bash
|
||||
# Build dan jalankan
|
||||
docker-compose up --build
|
||||
|
||||
# Hentikan services
|
||||
docker-compose down
|
||||
```
|
||||
|
||||
## Dokumentasi Terkait
|
||||
|
||||
- `Machine_Learning/README.md` — Dokumentasi workflow machine learning
|
||||
- `CLAUDE.md` — Panduan untuk Claude Code
|
||||
- `docker-compose.yml` — Konfigurasi Docker deployment
|
||||
- `apps/web/package.json` — Dependencies frontend
|
||||
- `apps/api/package.json` — Dependencies backend
|
||||
- `apps/ml-service/requirements.txt` — Dependencies ML service
|
||||
|
||||
## Kontribusi
|
||||
|
||||
1. Fork repository
|
||||
2. Buat branch fitur (`git checkout -b feature/amazing-feature`)
|
||||
3. Commit perubahan (`git commit -m 'Add amazing feature'`)
|
||||
4. Push ke branch (`git push origin feature/amazing-feature`)
|
||||
5. Buat Pull Request
|
||||
|
||||
## Lisensi
|
||||
|
||||
Distributed under the MIT License. See `LICENSE` for more information.
|
||||
|
||||
## Kontak
|
||||
|
||||
Asep Haryana Saputra - [GitHub](https://github.com/mytheclipse)
|
||||
|
||||
Project Link: [https://github.com/mytheclipse/ZeaVis-Edu](https://github.com/mytheclipse/ZeaVis-Edu)
|
||||
Reference in New Issue
Block a user