👁️ Computer Vision

🧠 How AI Sees Images

A Convolutional Neural Network (CNN) processes an image through layers:

• Input layer – pixels (e.g., 224×224×3 RGB values).
• Convolutional layers – detect edges, textures, shapes.
• Pooling layers – reduce size, keep important features.
• Fully connected layers – combine features to classify.

🚀 Live Demo: Image Classifier (MobileNet)

Upload any image or use a default one. The model (pre‑trained on 1,000+ object categories) will predict what it sees.

🔍 Classify Image 📷 Load Demo Image (Dog)

Upload an image and click "Classify".

// Load pre-trained MobileNet model
const model = await mobilenet.load();

// Get image element (or from file input)
const img = document.getElementById('preview');

// Classify
const predictions = await model.classify(img);
console.log(predictions); // [{className: "dog", probability: 0.92}, ...]

🌍 Real‑World Computer Vision

✨ Challenge: Test with Your Own Objects

Upload photos of everyday items (phone, book, coffee mug, fruit). Note which predictions are accurate and which are surprising. Try a photo with multiple objects – the model will return the top guess.