Zad1.zip -

In machine learning, a refers to the data representation extracted from the intermediate layers of a Deep Neural Network (DNN), such as a Convolutional Neural Network (CNN). Unlike "handcrafted" features (like edges or color histograms), deep features are automatically learned by the network and often capture complex, semantic information about the input. 2. Common Context for "zad1.zip"

: Reusing layers from a deep model to initialize a new task, where the "deep features" serve as the foundation for learning. zad1.zip

: Using a pre-trained model (like VGG16, ResNet, or AlexNet) to convert an image into a numerical vector (a "deep feature") for use in a simpler classifier like an SVM or k-Nearest Neighbors. In machine learning, a refers to the data

If you are working with Python (common for these tasks), deep features are typically extracted by removing the final classification layer of a model: Common Context for "zad1

import torch import torchvision.models as models # Load a pre-trained model model = models.resnet50(pretrained=True) # Remove the last fully connected layer to get features feature_extractor = torch.nn.Sequential(*(list(model.children())[:-1])) # 'output' will be the deep feature vector for an input image # output = feature_extractor(input_image) Use code with caution. Copied to clipboard

: Applying techniques like PCA or Autoencoders to compress high-dimensional deep features into a more manageable "compact feature vector".