Skip to main content

​
Overview

Pretrained models provide powerful feature extraction capabilities for anomaly detection without requiring domain-specific training. These models have been trained on large-scale datasets and can extract rich, hierarchical representations from images.

​
Available Models

ResNet-50 (CNN)

Classic convolutional neural network pretrained on ImageNet

CLIP

Vision-language model with zero-shot capabilities

​
Why Pretrained Models?

  1. No Domain-Specific Training Required: Features can be extracted immediately without training on the target dataset
  2. Rich Representations: Models learn hierarchical features from edges to complex patterns
  3. Transfer Learning: Knowledge from large datasets transfers to specialized domains
  4. Computational Efficiency: No expensive training phase required for feature extraction
Edit this page on GitHub or file an issue.
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.