Digit recognizer

Domains/AI-ML/MiniProjects/digit_recognizer/README.md

@@ -0,0 +1,61 @@
+                 MNIST DIGIT CLASSIFIER
+
+**Contributor:** Dhiraj201226
+**Domain:** AI-ML
+**Difficulty:** [Beginner]
+
+## Description
+It's a complete machine learning project, written in Python using the PyTorch library. Its entire purpose is to look at a small, grayscale image of a handwritten number and correctly identify what number it is (from 0 to 9 currently).
+
+This project solves the fundamental problem of automatic image classification.
+
+Specifically, it provides an automated solution to the real-world challenge of Optical Character Recognition (OCR) for digits. This is a classic problem in computer vision.
+
+## Dataset
+- **Source**: Automatically via torchvision.datasets.MNIST(..., download=True), which automatically downloads the dataset from official mirrors (like one hosted on AWS S3).
+
+- **Size**: 70000 images
+
+- **Features**: Each image is 28*28 pixel grayscale image
+and it is flattened into a 1-dimensional vector of 784 features (28 * 28 = 784).
+
+## Model Architecture
+- Algorithm/Architecture used:
+Simple Feed-Forward Neural Network (MLP) built with PyTorch (nn.Module). It has one hidden layer: Input (784 features) -> Linear(784 -> 512) -> ReLU -> Linear(512 -> 10) -> Output Scores.
+
+               KEY HYPERPARAMETERS
+
+ ->INPUT_SIZE: 784
+->NUM_CLASSES: 10
+->NUM_EPOCHS: 50
+->BATCH_SIZE: 100 
+->LEARNING_RATE: 0.001
+
+               - Training approach
+The model is trained on 50 epoches for better accuracy and also the 70k dataset is divided into 56k for training and 14k for test
+dataset is traine on 56k immages.
+
+## Requirements
+matplotlib==3.9.0
+numpy==1.26.4
+packaging==24.1
+pillow==10.4.0
+torch==2.3.1+cpu
+torchvision==0.18.1+cpu
+
+# Train model
+python train.py{
+    RUN model.py and config.py before it.
+}
+
+# Make predictions
+python predict.py
+
+## Results
+- Accuracy: 99.76%(training dataset)
+- Other metrics: final training loss:
+- Sample outputs:In random infernece it predicted all images correct. 
+refer to Figure_1.png
+## References
+->Learning representations by back-propagating errors (1986) - Rumelhart, Hinton, Williams
+->ImageNet Classification with Deep Convolutional Neural Networks (2012)- Krizhevsky, Sutskever, Hinton

Domains/AI-ML/MiniProjects/digit_recognizer/config.py

@@ -0,0 +1,20 @@
+# config.py
+import torch
+
+# --- Model Hyperparameters ---
+INPUT_SIZE = 784      # 28x28
+HIDDEN_SIZE = 512
+NUM_CLASSES = 10
+
+# --- Training Hyperparameters ---
+NUM_EPOCHS = 15
+BATCH_SIZE = 100
+LEARNING_RATE = 0.001
+DATA_DIR = './data'
+MODEL_PATH = 'mnist_nn1.pth' 
+
+# --- Data & Reproducibility ---
+RANDOM_SEED = 42
+
+# --- Device ---
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")

Domains/AI-ML/MiniProjects/digit_recognizer/model.py

@@ -0,0 +1,15 @@
+# model.py
+import torch.nn as nn
+
+class NeuralNetwork(nn.Module):
+  def __init__(self, input_size, hidden_size, num_classes):
+    super(NeuralNetwork, self).__init__()
+    self.l1 = nn.Linear(input_size, hidden_size)
+    self.relu = nn.ReLU()
+    self.l2 = nn.Linear(hidden_size, num_classes)
+
+  def forward(self, x):
+    out = self.l1(x)
+    out = self.relu(out)
+    out = self.l2(out)
+    return out

Domains/AI-ML/MiniProjects/digit_recognizer/predict.py

@@ -0,0 +1,67 @@
+# predict.py
+import torch
+import torchvision.transforms as transforms
+from torchvision.datasets import MNIST
+from torch.utils.data import DataLoader, random_split, ConcatDataset
+import matplotlib.pyplot as plt
+
+# Import from our custom files
+from model import NeuralNetwork
+import config as cfg
+
+def get_test_batch():
+    torch.manual_seed(cfg.RANDOM_SEED)
+
+    transform = transforms.ToTensor()
+
+    # Load and split data exactly as in train.py
+    train_dataset = MNIST(root=cfg.DATA_DIR, train=True, transform=transform, download=True)
+    test_dataset = MNIST(root=cfg.DATA_DIR, train=False, transform=transform, download=True)
+    combined_dataset = ConcatDataset([train_dataset, test_dataset])
+    train_size = int(0.8 * len(combined_dataset))
+    test_size = len(combined_dataset) - train_size
+    _, test_dataset = random_split(combined_dataset, [train_size, test_size])
+
+    # Create the test loader
+    test_loader = DataLoader(test_dataset, batch_size=cfg.BATCH_SIZE, shuffle=False)
+
+    # Get one batch
+    examples = iter(test_loader)
+    example_data, example_targets = next(examples)
+
+    return example_data, example_targets
+
+def main():
+    # --- 1. Load a batch of test data ---
+    example_data, example_targets = get_test_batch()
+
+    # --- 2. Initialize and Load Model ---
+    model = NeuralNetwork(cfg.INPUT_SIZE, cfg.HIDDEN_SIZE, cfg.NUM_CLASSES).to(cfg.DEVICE)
+    try:
+        model.load_state_dict(torch.load(cfg.MODEL_PATH, map_location=cfg.DEVICE))
+    except FileNotFoundError:
+        print(f"Error: Model file not found at {cfg.MODEL_PATH}")
+        print("Please run train.py first to train and save the model.")
+        return
+
+    model.eval() 
+
+    # --- 3. Make Predictions ---
+    with torch.no_grad():
+        example_images = example_data.reshape(-1, 28 * 28).to(cfg.DEVICE)
+        outputs = model(example_images)
+        _, preds = torch.max(outputs.data, 1)
+
+    # --- 4. Visualize Predictions ---
+    print("Plotting predictions for 6 test images...")
+    plt.figure(figsize=(10, 4))
+    for i in range(6):
+        plt.subplot(2, 3, i + 1)
+        plt.imshow(example_data[i][0], cmap='gray')
+        plt.title(f'Actual: {example_targets[i].item()}, Predicted: {preds[i].item()}')
+        plt.axis('off')
+    plt.tight_layout()
+    plt.show()
+
+if __name__ == "__main__":
+    main()

Domains/AI-ML/MiniProjects/digit_recognizer/requirements.txt

@@ -0,0 +1,6 @@
+matplotlib==3.9.0
+numpy==1.26.4
+packaging==24.1
+pillow==10.4.0
+torch==2.3.1+cpu
+torchvision==0.18.1+cpu

Domains/AI-ML/MiniProjects/digit_recognizer/train.py

@@ -0,0 +1,101 @@
+# train.py
+import torch
+import torch.nn as nn
+import torchvision.transforms as transforms
+from torchvision.datasets import MNIST
+from torch.utils.data import DataLoader, random_split, ConcatDataset
+
+# Import from our custom files
+from model import NeuralNetwork
+import config as cfg
+
+def get_data_loaders(batch_size, random_seed, data_dir):
+
+    torch.manual_seed(random_seed)
+
+    transform = transforms.ToTensor()
+
+    # Load original train and test datasets
+    train_dataset = MNIST(root=data_dir, train=True, transform=transform, download=True)
+    test_dataset = MNIST(root=data_dir, train=False, transform=transform, download=True)
+
+    # Combine them
+    combined_dataset = ConcatDataset([train_dataset, test_dataset])
+
+    # Create new 80/20 split
+    train_size = int(0.8 * len(combined_dataset))
+    test_size = len(combined_dataset) - train_size
+    train_dataset, test_dataset = random_split(combined_dataset, [train_size, test_size])
+
+    # Create DataLoaders
+    train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
+    test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)
+
+    return train_loader, test_loader
+
+def main():
+    # --- 1. Load Data ---
+    train_loader, test_loader = get_data_loaders(cfg.BATCH_SIZE, cfg.RANDOM_SEED, cfg.DATA_DIR)
+
+    # --- 2. Initialize Model, Loss, and Optimizer ---
+    model = NeuralNetwork(cfg.INPUT_SIZE, cfg.HIDDEN_SIZE, cfg.NUM_CLASSES).to(cfg.DEVICE)
+    criterion = nn.CrossEntropyLoss()
+    optimizer = torch.optim.Adam(model.parameters(), lr=cfg.LEARNING_RATE)
+
+    # --- 3. Training Loop ---
+    print("Starting training...")
+    n_total_steps = len(train_loader)
+    for epoch in range(cfg.NUM_EPOCHS):
+        for i, (images, labels) in enumerate(train_loader):
+            # Reshape images to (batch_size, 784) and move to device
+            images = images.reshape(-1, 28 * 28).to(cfg.DEVICE)
+            labels = labels.to(cfg.DEVICE)
+
+            # Forward pass
+            predict_outputs = model(images)
+            loss = criterion(predict_outputs, labels)
+
+            # Backward pass and optimization
+            loss.backward()
+            optimizer.step()
+            optimizer.zero_grad()
+
+            if (i + 1) % 500 == 0:
+                print(f'Epoch [{epoch + 1}/{cfg.NUM_EPOCHS}], Step [{i + 1}/{n_total_steps}], Loss: {loss.item():.4f}')
+
+    print("Training finished.")
+
+    # --- 4. Evaluation ---
+    with torch.no_grad():
+        train_correct = 0
+        train_samples = 0
+        for images, labels in train_loader:
+            images = images.reshape(-1, 28 * 28).to(cfg.DEVICE)
+            labels = labels.to(cfg.DEVICE)
+            outputs = model(images)
+            _, predicted = torch.max(outputs.data, 1)
+            train_samples += labels.size(0)
+            train_correct += (predicted == labels).sum().item()
+
+        train_acc = 100.0 * train_correct / train_samples
+        print(f'Accuracy of the network on the {train_samples} training images: {train_acc:.2f} %')
+
+        test_correct = 0
+        test_samples = 0
+        for images, labels in test_loader:
+            images = images.reshape(-1, 28 * 28).to(cfg.DEVICE)
+            labels = labels.to(cfg.DEVICE)
+            outputs = model(images)
+            _, predicted = torch.max(outputs.data, 1)
+            test_samples += labels.size(0)
+            test_correct += (predicted == labels).sum().item()
+
+        test_acc = 100.0 * test_correct / test_samples
+        print(f'Accuracy of the network on the {test_samples} test images: {test_acc:.2f} %')
+
+    # --- 5. Save the Model ---
+    torch.save(model.state_dict(), cfg.MODEL_PATH)
+    print(f"Model saved to {cfg.MODEL_PATH}")
+
+if __name__ == "__main__":
+    main()

Domains/Frontend/MiniProjects/TodoApp/index.html

@@ -9,13 +9,11 @@
 </head>
 <body>
     <div class="container">
-        <!-- Header -->
         <header class="header">
             <h1>📝 My Todo List</h1>
             <p class="subtitle">Stay organized, stay productive</p>
         </header>
 
-        <!-- Input Section -->
         <div class="input-section">
             <input 
                 type="text" 
@@ -30,7 +28,6 @@ <h1>📝 My Todo List</h1>
             </button>
         </div>
 
-        <!-- Filter Section -->
         <div class="filter-section">
             <div class="filter-buttons">
                 <button class="filter-btn active" data-filter="all">All</button>
@@ -42,26 +39,21 @@ <h1>📝 My Todo List</h1>
             </div>
         </div>
 
-        <!-- Tasks List -->
         <ul id="taskList" class="task-list">
-            <!-- Tasks will be dynamically added here -->
         </ul>
 
-        <!-- Footer Actions -->
         <div class="footer-actions">
             <button id="clearCompleted" class="clear-btn">
                 Clear Completed
             </button>
         </div>
 
-        <!-- Empty State -->
         <div id="emptyState" class="empty-state">
             <div class="empty-icon">📋</div>
             <p>No tasks yet. Add one above to get started!</p>
         </div>
     </div>
 
-    <!-- Footer -->
     <footer class="app-footer">
         <p>Made with ❤️ for ProjectHive | <a href="https://github.com" target="_blank">View on GitHub</a></p>
     </footer>