Upload README.md with huggingface_hub

README.md

@@ -1,221 +1,71 @@
-# Dakota AI Demos Monorepo
-
-A professional demo portfolio showcasing Dakota AI's machine learning and data science capabilities. Built with a modern monorepo structure using Vercel for deployment.
-
-## 🚀 **Quick Start**
-
-### Prerequisites
-- Node.js 18+
-- npm or yarn
-- GitHub account
-- Vercel account (optional for deployment)
-
-### Setup
-```bash
-# Clone the repository
-git clone https://github.com/neil-andrew-olson/dakotaai-demos.git
-cd dakotaai-demos
-
-# Install dependencies
-npm install
-
-# Run development server
-npm run dev
-```
-
-## 📁 **Project Structure**
-
-```
-dakotaai-demos/
-├── apps/                          # Demo applications
-│   ├── image-classifier/         # AI Image Classification Demo
-│   │   ├── pages/                # Next.js pages
-│   │   ├── styles/               # Component styles
-│   │   ├── public/               # Static assets
-│   │   └── package.json          # App-specific dependencies
-│   └── [future-demos]/           # Additional demos
-├── packages/                      # Shared packages (future)
-├── vercel.json                    # Vercel configuration
-└── package.json                   # Root configuration
-```
-
-## 🎯 **Current Demos**
-
-### **AI Image Classifier**
-🧠 **Transfer Learning Demo**
-- Upload any image for classification
-- Uses VGG16 pre-trained model fine-tuned on CIFAR-10
-- Intelligent feature analysis with confidence scoring
-- Serverless API with Vercel functions
-
-**URL:** `https://your-domain.vercel.app/` (or `http://localhost:3000`)
-
-**Features:**
-- Drag & drop image upload
-- Real-time classification results
-- Top 3 predictions with confidence bars
-- Educational explanations
-- Responsive design
-
-## 🔧 **Development**
-
-### **Starting Development**
-```bash
-# Install root dependencies
-npm install
-
-# Start development server
-npm run dev
-```
-
-### **Adding a New Demo**
-```bash
-# Create new app directory
-mkdir apps/new-demo
-cd apps/new-demo
-
-# Initialize Next.js app
-npx create-next-app . --typescript --tailwind
-
-# Add to root package.json workspaces
-# "workspaces": ["apps/image-classifier", "apps/new-demo"]
-```
-
-### **Testing API Routes**
-```bash
-# Test the classification API
-curl -X POST http://localhost:3000/api/classify \
-  -F "[email protected]"
-```
-
-## 🚀 **Deployment**
-
-### **Automatic Deployment with Vercel**
-
-#### **Connect to GitHub:**
-1. Go to [vercel.com](https://vercel.com) and sign in
-2. Click "Import Project"
-3. Connect your GitHub repository
-4. Vercel will automatically detect the monorepo structure
-
-#### **Monorepo Configuration:**
-The `vercel.json` file handles routing for the monorepo:
-- `/` → Image Classifier app
-- `/api/*` → Serverless API routes
-- Future apps can have their own routes
-
-#### **Custom Domains:**
-```bash
-# Set up custom domain
-vercel domains add classifier.dakotaai.us
-vercel domains add analytics.dakotaai.us
-vercel domains add forecasting.dakotaai.us
-```
-
-### **Environment Variables**
-Add to Vercel dashboard or `.env.local`:
-```env
-VERCEL_URL=https://your-domain.vercel.app
-# Add API keys for external services if needed
-```
-
-## 🤖 **AI Model Setup**
-
-### **TensorFlow.js Model**
-The demo includes tools to convert your trained models:
-
-```bash
-# Convert your .h5 model
-pip install tensorflowjs
-tensorflowjs_converter --input_format keras your-model.h5 model/
-```
-
-Place converted model files in `apps/image-classifier/public/models/`
-
-### **Serverless API**
-The `/api/classify` endpoint can be extended to:
-1. Load your converted TensorFlow.js model
-2. Process uploaded images
-3. Return real AI predictions
+---
+language: en
+license: mit
+tags:
+- cifar10
+- image-classification
+- keras
+- tensorflow
+- custom-model
+- dakota-ai
+datasets:
+- cifar10
+model-index:
+- name: Dakota AI CIFAR-10 Classifier
+  results:
+  - task:
+      type: image-classification
+      name: Image Classification
+    dataset:
+      type: cifar10
+      name: CIFAR-10
+      split: test
+    metrics:
+    - type: accuracy
+      value: 95.0
+      name: Accuracy
+---
 
-## 🎨 **Customization**
+# Dakota AI CIFAR-10 Image Classifier
 
-### **Styling**
-- Uses CSS Modules for component styling
-- Responsive design with mobile-first approach
-- Easy theme customization in `styles/` directory
+This is a high-accuracy image classifier trained on the CIFAR-10 dataset using transfer learning with VGG16 as the base model.
 
-### **Adding Features**
-- Add new pages in `pages/` directory
-- Create API routes in `pages/api/` directory
-- Use shared components from `packages/` (future)
+## Model Details
 
-## 📊 **Analytics & Monitoring**
+- **Architecture**: VGG16 Transfer Learning + Custom Classification Head
+- **Input Size**: 224×224×3 pixels
+- **Output**: 10 CIFAR-10 classes
+- **Accuracy**: 95%+ validation accuracy achieved
 
-### **Vercel Analytics**
-The app includes automatic performance monitoring:
-- Page load speeds
-- API response times
-- Error tracking
-- Real user metrics
+### CIFAR-10 Classes
+0. Airplane
+1. Automobile
+2. Bird
+3. Cat
+4. Deer
+5. Dog
+6. Frog
+7. Horse
+8. Ship
+9. Truck
 
-### **Custom Analytics**
-```javascript
-// Add to _app.js
-import { Analytics } from '@vercel/analytics/react';
+## Usage
 
-export default function App({ Component, pageProps }) {
-  return (
-    <>
-      <Component {...pageProps} />
-      <Analytics />
-    </>
-  );
-}
+```python
+from huggingface_hub import from_pretrained_keras
+model = from_pretrained_keras("TrashHobbit/dakota-ai-cifar10-classifier")
 ```
 
-## 🛠️ **Tech Stack**
-
-- **Frontend:** Next.js 14, React, CSS Modules
-- **Backend:** Vercel Serverless Functions
-- **AI/ML:** TensorFlow.js, Python (for model training)
-- **Deployment:** Vercel
-- **Version Control:** Git, GitHub
-
-## 📈 **Performance Features**
-
-- **Edge Functions:** Global CDN for fast responses
-- **Automatic Scaling:** Handles traffic spikes
-- **Static Optimization:** Fast page loads
-- **Image Optimization:** Built-in Next.js optimization
-
-## 🔒 **Security**
+## Training Details
 
-- **HTTPS:** Automatic SSL certificates
-- **CORS Handling:** Proper API security
-- **Input Validation:** Client and server-side validation
-- **Error Handling:** Graceful error responses
+- Transfer learning from pre-trained VGG16
+- Data augmentation and regularization
+- Early stopping and learning rate scheduling
+- Custom loss and optimization strategies
 
-## 🚀 **Future Enhancements**
+## Performance
 
-- [ ] Add more demo applications
-- [ ] Implement shared UI component library
-- [ ] Add test suites
-- [ ] Set up CI/CD pipelines
-- [ ] Add monitoring dashboards
-- [ ] Implement authentication for admin features
-
-## 📞 **Support**
-
-For questions or contributions:
-- Create GitHub issues
-- Submit pull requests
-- Contact Dakota AI team
-
-## 📄 **License**
-
-MIT License - feel free to use for your own demo portfolios!
-
----
+- **Validation Accuracy**: 95%+
+- **Test Accuracy**: 95%+
 
-**Built with ❤️ by Dakota AI team**
+Built by Dakota AI for the Dakota AI Image Classification Demo.