Deploy SD API

Dockerfile

@@ -0,0 +1,28 @@
+FROM python:3.10-slim
+
+ENV PYTHONUNBUFFERED=1
+
+WORKDIR /app
+
+COPY api_server.py /app/
+
+RUN apt-get update && apt-get install -y git && rm -rf /var/lib/apt/lists/*
+
+RUN pip install --no-cache-dir \
+    torch torchvision torchaudio \
+    --index-url https://download.pytorch.org/whl/cpu
+
+RUN pip install --no-cache-dir \
+    diffusers \
+    transformers \
+    safetensors \
+    peft \
+    flask \
+    pillow \
+    accelerate
+
+RUN mkdir -p /app/model /app/generated_images
+
+EXPOSE 7860
+
+CMD ["python3", "-u", "/app/api_server.py"]

api_server.py

@@ -0,0 +1,128 @@
+#!/usr/bin/env python3
+"""
+API Flask pour générer des images
+Endpoint unique: /generate
+"""
+
+import torch
+from pathlib import Path
+from flask import Flask, request, jsonify, send_file
+from diffusers import StableDiffusionPipeline
+from datetime import datetime
+import io
+
+app = Flask(__name__)
+
+# Configuration
+OUTPUT_DIR = Path("/app/generated_images")
+OUTPUT_DIR.mkdir(exist_ok=True, parents=True)
+
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+
+# Variable globale
+pipeline = None
+model_loaded = False
+
+def load_model():
+    """Charge le modèle au démarrage"""
+    global pipeline, model_loaded
+    
+    print("\n" + "="*70)
+    print("🤖 Chargement du modèle fusionné...")
+    print("="*70 + "\n")
+    
+    try:
+        print(f"📱 Appareil: {DEVICE}")
+        
+        dtype = torch.float32 if DEVICE == "cpu" else torch.float16
+        
+        pipeline = StableDiffusionPipeline.from_pretrained(
+            "K2MAR/mon-modele-sd",
+            torch_dtype=dtype,
+            safety_checker=None
+        ).to(DEVICE)
+        
+        pipeline.enable_attention_slicing()
+        
+        model_loaded = True
+        print("="*70)
+        print("✅ Modèle prêt!")
+        print("="*70 + "\n")
+        return True
+    except Exception as e:
+        print(f"❌ Erreur: {e}")
+        return False
+
+@app.route('/health', methods=['GET'])
+def health():
+    return jsonify({
+        "status": "ok" if model_loaded else "loading",
+        "device": DEVICE,
+        "model_loaded": model_loaded
+    })
+
+@app.route('/generate', methods=['POST'])
+def generate():
+    if not model_loaded:
+        return jsonify({"error": "Model not loaded"}), 503
+    
+    try:
+        data = request.get_json()
+        
+        if not data or "prompt" not in data:
+            return jsonify({"error": "Missing 'prompt' in request"}), 400
+        
+        prompt = data.get("prompt", "")
+        steps = int(data.get("steps", 30))
+        guidance_scale = float(data.get("guidance_scale", 7.5))
+        
+        if not prompt:
+            return jsonify({"error": "Prompt cannot be empty"}), 400
+        
+        if steps < 1 or steps > 50:
+            return jsonify({"error": "Steps must be 1-50"}), 400
+        
+        print(f"\n🎨 Génération: {prompt}")
+        
+        with torch.no_grad():
+            image = pipeline(
+                prompt,
+                num_inference_steps=steps,
+                guidance_scale=guidance_scale,
+                height=512,
+                width=512
+            ).images[0]
+        
+        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+        filepath = OUTPUT_DIR / f"generated_{timestamp}.png"
+        image.save(filepath)
+        
+        img_io = io.BytesIO()
+        image.save(img_io, 'PNG')
+        img_io.seek(0)
+        
+        return send_file(img_io, mimetype='image/png')
+    
+    except Exception as e:
+        print(f"❌ Erreur: {str(e)}\n")
+        return jsonify({"error": str(e)}), 500
+
+@app.route('/', methods=['GET'])
+def home():
+    return jsonify({
+        "service": "LoRA Solar Panel Generator API",
+        "version": "1.0",
+        "device": DEVICE,
+        "model_loaded": model_loaded,
+        "endpoints": {
+            "health": "GET /health",
+            "generate": "POST /generate"
+        }
+    })
+
+if __name__ == '__main__':
+    if not load_model():
+        exit(1)
+    
+    print("\n🚀 Serveur démarrage sur 0.0.0.0:7860\n")
+    app.run(host='0.0.0.0', port=7860, debug=False, threaded=True)