Update index.js

index.js

@@ -1,76 +1,290 @@
-import { pipeline } from 'https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.7.6';
+import { AutoProcessor, CLIPVisionModelWithProjection, RawImage, env } from 'https://cdn.jsdelivr.net/npm/@xenova/transformers@2.6.0';
+
+// Since we will download the model from the Hugging Face Hub, we can skip the local model check
+env.allowLocalModels = false;
 
 // Reference the elements that we will need
-const status = document.getElementById('status');
-const fileUpload = document.getElementById('upload');
-const imageContainer = document.getElementById('container');
-const example = document.getElementById('example');
+const statusText = document.getElementById('status-text');
+const fileUpload = document.getElementById('file-upload');
+const dropZone = document.getElementById('drop-zone');
+const imagePreview1 = document.getElementById('image-preview-1');
+const imagePreview2 = document.getElementById('image-preview-2');
+const meterContainer = document.getElementById('meter-container');
+const spinner = document.querySelector('.spinner');
+const showGraphBtn = document.getElementById('show-graph-btn');
+const graphModal = document.getElementById('graph-modal');
+const closeModalBtn = document.querySelector('.close-button');
+const resetZoomBtn = document.getElementById('reset-zoom-btn');
+const graphContainerModal = document.getElementById('graph-container-modal');
 
-const EXAMPLE_URL = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/city-streets.jpg';
+// Load processor and vision model for more direct embedding control
+statusText.textContent = 'Loading model...';
+spinner.style.display = 'block';
+const processor = await AutoProcessor.from_pretrained('Xenova/clip-vit-base-patch16');
+const vision_model = await CLIPVisionModelWithProjection.from_pretrained('Xenova/clip-vit-base-patch16');
+statusText.textContent = 'Ready';
+spinner.style.display = 'none';
 
-// Create a new object detection pipeline
-status.textContent = 'Loading model...';
-const detector = await pipeline('object-detection', 'Xenova/detr-resnet-50');
-status.textContent = 'Ready';
+let imageSrc1 = null;
+let imageSrc2 = null;
+let lastEmbeds = null;
 
-example.addEventListener('click', (e) => {
-    e.preventDefault();
-    detect(EXAMPLE_URL);
+// Initial setup of upload placeholders
+clearUploads();
+
+
+// Prevent default drag behaviors
+['dragenter', 'dragover', 'dragleave', 'drop'].forEach(eventName => {
+    dropZone.addEventListener(eventName, preventDefaults, false);
+    document.body.addEventListener(eventName, preventDefaults, false);
 });
 
-fileUpload.addEventListener('change', function (e) {
-    const file = e.target.files[0];
-    if (!file) {
+// Highlight drop zone when item is dragged over it
+['dragenter', 'dragover'].forEach(eventName => {
+    dropZone.addEventListener(eventName, () => dropZone.classList.add('highlight'), false);
+});
+
+['dragleave', 'drop'].forEach(eventName => {
+    dropZone.addEventListener(eventName, () => dropZone.classList.remove('highlight'), false);
+});
+
+// Handle dropped files
+dropZone.addEventListener('drop', handleDrop, false);
+
+// Handle clear button click
+const clearBtn = document.getElementById('clear-btn');
+clearBtn.addEventListener('click', clearUploads);
+
+// Handle file selection via click
+fileUpload.addEventListener('change', handleFileSelect);
+
+// Modal event listeners
+showGraphBtn.addEventListener('click', () => {
+    if (lastEmbeds) {
+        graphModal.style.display = 'block';
+        renderEmbeddingGraph(lastEmbeds.embeds1, lastEmbeds.embeds2);
+    }
+});
+
+closeModalBtn.addEventListener('click', () => {
+    graphModal.style.display = 'none';
+});
+
+window.addEventListener('click', (event) => {
+    if (event.target == graphModal) {
+        graphModal.style.display = 'none';
+    }
+});
+
+function preventDefaults(e) {
+    e.preventDefault();
+    e.stopPropagation();
+}
+
+function handleDrop(e) {
+    const dt = e.dataTransfer;
+    const files = dt.files;
+    handleFiles(files);
+}
+
+function handleFileSelect(e) {
+    handleFiles(e.target.files);
+}
+
+function handleFiles(files) {
+    const filesArray = Array.from(files);
+    if (filesArray.length === 0) {
         return;
     }
 
+    // If no image is uploaded yet, fill the first slot. Otherwise, fill the second.
+    if (!imageSrc1) {
+        handleIndividualFile(filesArray[0], '1');
+        if (filesArray.length > 1) {
+            handleIndividualFile(filesArray[1], '2');
+        }
+    } else {
+        handleIndividualFile(filesArray[0], '2');
+    }
+}
+
+function handleIndividualFile(file, target) {
+    if (!file) {
+        return;
+    }
+    
     const reader = new FileReader();
+    reader.onload = function (e2) {
+        const imageSrc = e2.target.result;
 
-    // Set up a callback when the file is loaded
-    reader.onload = e2 => detect(e2.target.result);
+        if (target === '1') {
+            imageSrc1 = imageSrc;
+            imagePreview1.innerHTML = `<img src="${imageSrc1}" alt="uploaded image 1">`;
+        } else if (target === '2') {
+            imageSrc2 = imageSrc;
+            imagePreview2.innerHTML = `<img src="${imageSrc2}" alt="uploaded image 2">`;
+        }
 
+        checkAndCompare();
+    };
     reader.readAsDataURL(file);
-});
+}
+
+function checkAndCompare() {
+    if (imageSrc1 && imageSrc2) {
+        compareImages(imageSrc1, imageSrc2);
+    }
+}
 
+function clearUploads() {
+    const placeholder = `<div class="placeholder">
+        <i class="fas fa-image"></i>
+        <p>Image preview</p>
+    </div>`;
+    imagePreview1.innerHTML = placeholder;
+    imagePreview2.innerHTML = placeholder;
 
-// Detect objects in the image
-async function detect(img) {
-    imageContainer.innerHTML = '';
-    imageContainer.style.backgroundImage = `url(${img})`;
+    imageSrc1 = null;
+    imageSrc2 = null;
 
-    status.textContent = 'Analysing...';
-    const output = await detector(img, {
-        threshold: 0.5,
-        percentage: true,
-    });
-    status.textContent = '';
-    output.forEach(renderBox);
+    lastEmbeds = null;
+    showGraphBtn.style.display = 'none';
+    // Reset file input
+    fileUpload.value = '';
+
+    meterContainer.innerHTML = '';
+}
+
+// Function to calculate cosine similarity between two vectors
+function cosineSimilarity(vecA, vecB) {
+    let dotProduct = 0.0;
+    let normA = 0.0;
+    let normB = 0.0;
+
+    for (let i = 0; i < vecA.length; i++) {
+        dotProduct += vecA[i] * vecB[i];
+        normA += vecA[i] * vecA[i];
+        normB += vecB[i] * vecB[i];
+    }
+
+    return dotProduct / (Math.sqrt(normA) * Math.sqrt(normB));
+}
+
+// Compare the two images using direct embedding calculation
+async function compareImages(img1, img2) {
+    statusText.textContent = 'Extracting embeddings...';
+    spinner.style.display = 'block';
+
+    try {
+        // Load images using RawImage
+        const image1 = await RawImage.read(img1);
+        const image2 = await RawImage.read(img2);
+
+        // Process images and compute embeddings
+        const image_inputs1 = await processor(image1);
+        const image_inputs2 = await processor(image2);
+
+        const { image_embeds: embeds1 } = await vision_model(image_inputs1);
+        const { image_embeds: embeds2 } = await vision_model(image_inputs2);
+
+        // Calculate cosine similarity
+        const similarity = cosineSimilarity(embeds1.data, embeds2.data);
+        lastEmbeds = { embeds1: embeds1.data, embeds2: embeds2.data };
+
+        statusText.textContent = 'Ready';
+        spinner.style.display = 'none';
+        renderResults(similarity);
+    } catch (error) {
+        statusText.textContent = '';
+        spinner.style.display = 'none';
+        meterContainer.innerHTML = `<div class="error"><p>Failed to compare images: ${error.message}</p></div>`;
+        console.error('Comparison error:', error);
+    }
+}
+
+// Render the comparison results
+function renderResults(similarity) {
+    meterContainer.innerHTML = '';
+
+    // Show the button
+    showGraphBtn.style.display = 'block';
+
+    // Create similarity meter
+    const meterElement = document.createElement('div'); // This will be a wrapper
+    meterElement.className = 'similarity-meter';
+
+    const score = Math.round(similarity * 100);
+    const meterValue = Math.max(0, Math.min(100, score));
+
+    meterElement.innerHTML = `
+        <div class="meter-label">Similarity Score: ${meterValue}%</div>
+        <div class="meter-container">
+            <div class="meter-bar" style="width: ${meterValue}%"></div>
+        </div>
+        <div class="meter-description">${getSimilarityDescription(similarity)}</div>
+    `;
+
+    meterContainer.appendChild(meterElement);
 }
 
-// Render a bounding box and label on the image
-function renderBox({ box, label }) {
-    const { xmax, xmin, ymax, ymin } = box;
-
-    // Generate a random color for the box
-    const color = '#' + Math.floor(Math.random() * 0xFFFFFF).toString(16).padStart(6, 0);
-
-    // Draw the box
-    const boxElement = document.createElement('div');
-    boxElement.className = 'bounding-box';
-    Object.assign(boxElement.style, {
-        borderColor: color,
-        left: 100 * xmin + '%',
-        top: 100 * ymin + '%',
-        width: 100 * (xmax - xmin) + '%',
-        height: 100 * (ymax - ymin) + '%',
-    })
-
-    // Draw label
-    const labelElement = document.createElement('span');
-    labelElement.textContent = label;
-    labelElement.className = 'bounding-box-label';
-    labelElement.style.backgroundColor = color;
-
-    boxElement.appendChild(labelElement);
-    imageContainer.appendChild(boxElement);
+function renderEmbeddingGraph(embeds1, embeds2) {
+    graphContainerModal.innerHTML = `
+        <h3 class="graph-title">Embedding Visualization</h3>
+        <canvas id="embedding-chart"></canvas>
+    `;
+
+    const ctx = document.getElementById('embedding-chart').getContext('2d');
+    new Chart(ctx, {
+        type: 'line',
+        data: {
+            labels: Array.from({ length: embeds1.length }, (_, i) => i), // Dimension index
+            datasets: [{
+                label: 'Image 1 Embedding',
+                data: embeds1,
+                borderColor: 'rgba(0, 123, 255, 0.8)',
+                backgroundColor: 'rgba(0, 123, 255, 0.1)',
+                borderWidth: 1,
+                pointRadius: 0,
+            }, {
+                label: 'Image 2 Embedding',
+                data: embeds2,
+                borderColor: 'rgba(40, 167, 69, 0.8)',
+                backgroundColor: 'rgba(40, 167, 69, 0.1)',
+                borderWidth: 1,
+                pointRadius: 0,
+            }]
+        },
+        options: {
+            responsive: true,
+            plugins: {
+                legend: { position: 'top' },
+                zoom: {
+                    pan: {
+                        enabled: true,
+                        mode: 'xy',
+                        modifierKey: null, // Allow panning without holding a key
+                    },
+                    zoom: {
+                        wheel: {
+                            enabled: true,
+                        },
+                        pinch: {
+                            enabled: true
+                        },
+                        mode: 'xy',
+                    }
+                }
+            }
+        }
+    });
 }
+
+function getSimilarityDescription(similarity) {
+    if (similarity > 0.9) {
+        return "🔥 Extremely similar - These images are nearly identical!";
+    } else if (similarity > 0.7) {
+        return "👍 Very similar - These images share strong visual characteristics.";
+    } else {
+        return "🚫 Not similar - These images appear to be very different.";
+    }
+}