python-server/QUICKSTART.md

🚀 Quick Start: FREE AI Alt Text Generation

2-Minute Setup (100% FREE!)

Step 1: Install Dependencies

cd "Cycle 2 Testing/Accessibility-Checker-BE/python-server"
pip install -r requirements.txt

That's it! No configuration needed. The system works with smart defaults.

First run note: The AI model downloads ~1-2GB (one time only, then cached)

Step 2: Start the Server

python server2.py

Look for: ✅ Local AI vision model loaded (BLIP - 100% FREE, No Costs)

Step 3: Test It!

Upload a PowerPoint through the frontend. The system will:

• ✅ Analyze accessibility issues
• ✅ Generate AI alt text for images using FREE local AI
• ✅ Create a remediated file for download
• ✅ Zero API costs, zero API keys needed!

Optional: Customize Settings

If you want to change settings (like using a different AI model):

cp .env.example .env
# Edit .env with any text editor to customize

But don't worry - the system works perfectly without .env! It's completely optional.

---

What's New?

Before (Placeholder Alt Text)

"Image on slide 3"
"decorative"

After (FREE AI-Generated Alt Text)

"Bar chart with four colored bars showing increasing values"
"Person standing at whiteboard presenting to seated audience"
"Company logo with red and blue colors"

---

How It Works

🆓 The Only Option: Local BLIP Model (100% FREE!)

Local BLIP AI Model

• ✅ 100% Free, unlimited usage
• ✅ Runs on your computer (offline after first download)
• ✅ No internet required for processing
• ✅ No API keys needed
• ✅ No account creation
• ✅ No surprise billing - ever!
• ✅ Fast and good quality (7/10)
• ⬇️ ~1GB download on first run
• ⚡ Instant on subsequent runs

Why This Setup?

All OpenAI references have been completely removed from the project to eliminate any possibility of surprise billing. The free local AI model is:

• Good enough - Works great for academic projects
• Cost effective - $0 per image vs $0.17 with paid APIs
• Simple - No configuration needed
• Safe - Runs on your own computer, no data sent anywhere

---

Configuration (100% Optional!)

Why no .env file is needed

The system works perfectly with smart defaults:

• ✅ Uses local BLIP model automatically
• ✅ Enables AI alt text generation
• ✅ No API keys to configure

Just install and run - that's it!

Optional: Customize (Create .env)

If you want to change settings, copy the template:

# Copy template
cp .env.example .env

# Edit with your preferred editor
# Optional settings you might change:
LOCAL_VISION_MODEL=blip-base      # Use blip-large for better quality
ENABLE_AI_ALT_TEXT=true           # Set to false to disable AI (for debugging)

See ENV_FILE_GUIDE.md for complete .env documentation.

---

Server Console Output

When everything is working:

✅ Local AI vision model loaded (BLIP - 100% FREE, No Costs)
🚀 Starting alt text remediation for: document.pptx
   AI Mode: LOCAL (100% FREE - No Costs)
🤖 Using FREE local AI (BLIP) for slide 1
  ✅ AI generated alt text for Picture 1: 'Professional man in business suit...'
✅ Remediation complete: 3 images processed
   🤖 3 alt texts generated by FREE local AI (no cost)

---

Troubleshooting

Problem: Slow download on first run

Explanation: System is downloading BLIP AI model (~1-2GB) Solution: This only happens once. Subsequent runs are instant. Be patient! Time estimate: 5-15 minutes depending on internet

Problem: "transformers not installed"

Solution:

pip install -r requirements.txt

Problem: "ModuleNotFoundError: No module named 'local_vision'"

Solution: Make sure you're running from the python-server/ directory

cd python-server
python server2.py

Problem: Out of memory errors

Solution: Close other programs or use smaller model

# In .env:
LOCAL_VISION_MODEL=blip-base

Problem: Alt text not being generated

Check the console output:

1. Does it show "✅ Local AI vision model loaded"?
2. Are images in supported formats (PNG, JPG, GIF)?
3. Is ENABLE_AI_ALT_TEXT set to true?

Run diagnostics:

python test_ai_setup.py

Problem: "This model requires transformers version X.X"

Solution:

pip install --upgrade transformers torch

---

Cost: FREE Forever!

Item	Cost
Local BLIP AI Model	$0
First download (one-time)	$0
Unlimited alt text generation	$0
Monthly hosting	$0 (free tier)
Total for entire team	$0 forever

Compared to alternatives:

• OpenAI: ~$0.17/image = $5-10 per presentation
• Google Vision: $1.50/100 images
• Azure: $1/$5/10 per 1000 requests
• Our solution: $0 per anything! 🎉

---

Documentation

For more detailed information, see:

• ENV_FILE_GUIDE.md - Complete .env explanation (optional)
• OPENAI_REMOVAL_COMPLETE.md - Why OpenAI was removed for safety
• AI_ALT_TEXT_SETUP.md - Deep technical documentation
• STUDENT_SETUP.md - Student-friendly setup guide
• FREE_AI_OPTIONS.md - Comparison of all free alternatives

---

Summary

✅ Fastest Setup:

pip install -r requirements.txt
python server2.py

✅ No Configuration Needed: Works with defaults

✅ 100% FREE: No API keys, no monthly bills, no surprises

✅ High Quality: BLIP model produces excellent alt text descriptions

✅ Easy to Use: Upload PowerPoint, download fixed version

✅ For Students: Zero cost, zero complexity

Ready to generate alt text for your presentations! 🚀