A06

Download this page as PDF

Dataset Preparation

• Use torchvision.datasets.VOCDetection to load the Pascal VOC 2007 or 2012 dataset.
• Extract image-level labels from annotations.
• Convert the dataset into PyTorch tensors.
• Split the dataset into training and validation sets (e.g., 80% training, 20% validation).
• Resize and normalize the image data.

Data Visualization

• Display at least 5 sample images with their bounding boxes and labels.
• Plot a bar chart showing the frequency of each object class.
• Plot a pie chart of the top 5 most common classes.

Build the Classification Model

• Choose a pre-trained model from torchvision.models (e.g., ResNet, VGG).
• Replace the final layer with a new fully connected layer to output predictions for 20 classes.
• Use ReLU activation in hidden layers and softmax (implicitly handled by loss function) for the output.

Train the Model

• Define a suitable loss function (e.g., CrossEntropyLoss).
• Choose an optimizer (e.g., Adam or SGD).
• Train the model for a fixed number of epochs (e.g., 10–20 epochs).
• Track and display training and validation loss per epoch.

Evaluate the Model

• Compute accuracy, precision, and recall on the validation set.
• Display a confusion matrix to visualize classification performance.