Deaf Connect AI — Deployment Guide

Project Structure

deaf_connect_api/
├── main.py           ← FastAPI backend
├── labels.py         ← 16 sign labels
├── requirements.txt  ← Python dependencies
├── render.yaml       ← Render config
├── model/
│   └── sign_model.h5 ← Your trained model (copy here)
└── static/
    └── api-client.js ← Frontend JS client

---

STEP 1 — Prepare Files

Copy your trained model:

model/sign_model.h5  ← apna trained model yahan rakho
labels.py            ← already ready hai

---

STEP 2 — GitHub Pe Push Karo

cd deaf_connect_api
git init
git add .
git commit -m "Deaf Connect AI FastAPI backend"
git branch -M main
git remote add origin https://github.com/YOUR_USERNAME/deaf-connect-api.git
git push -u origin main

---

STEP 3 — Render Pe Deploy Karo

1. render.com pe jao → Sign up (free)
2. "New +" → "Web Service"
3. GitHub repo connect karo
4. Settings:
   • Name: deaf-connect-api
   • Environment: Python 3
   • Build Command: pip install -r requirements.txt
   • Start Command: uvicorn main:app --host 0.0.0.0 --port $PORT
5. "Create Web Service" click karo
6. Wait 3-5 minutes — deploy hoga
7. URL milega: https://deaf-connect-api.onrender.com

---

STEP 4 — Frontend Update Karo (Vercel)

api-client.js mein yeh line update karo:

const API_BASE_URL = "https://deaf-connect-api.onrender.com";

Phir Vercel pe push karo — auto deploy ho jayega.

---

API Endpoints

Method	URL	Description
GET	/health	Server status check
GET	/labels	All 16 labels
GET	/sentences	All 10 sentences
POST	/predict	Landmarks → Sign label
POST	/build-sentence	Words → Sentence
POST	/speak	Text → Audio (base64)
POST	/transcribe	Audio → Text

---

Test API (after deploy)

# Health check
curl https://deaf-connect-api.onrender.com/health

# Get labels
curl https://deaf-connect-api.onrender.com/labels

# Test predict (dummy data)
curl -X POST https://deaf-connect-api.onrender.com/predict \
  -H "Content-Type: application/json" \
  -d '{"landmarks": [0.5]*126, "confidence_threshold": 0.65}'

# Test sentence
curl -X POST https://deaf-connect-api.onrender.com/build-sentence \
  -H "Content-Type: application/json" \
  -d '{"words": ["accident", "now", "help"]}'

---

API Docs (Auto-generated)

After deploy, visit:

• Swagger UI: https://deaf-connect-api.onrender.com/docs
• ReDoc: https://deaf-connect-api.onrender.com/redoc

---

Important Notes

1. Free Render — server sleeps after 15 min inactivity

   • First request after sleep takes ~30 seconds (cold start)
   • Add a health ping every 10 min to keep alive

2. Model size — sign_model.h5 large hai toh GitHub LFS use karo:

   git lfs install
   git lfs track "*.h5"
   git add .gitattributes

3. pyttsx3 on Linux — Render Linux pe hai, pyttsx3 needs espeak: Add this file: build.sh

   apt-get install -y espeak
   pip install -r requirements.txt

   Change build command in Render to: bash build.sh