分类模型检测开源项目

# Classification Post-Processing for Hawk

## Overview

A generic ONNX classification block that runs after anomaly detection in Hawk's
post-processing pipeline. It classifies each anomaly detection crop to determine
if it is a real defect or a false positive (overkill).

One script, many models. Each camera/pipeline points to its own model folder.

## Folder Structure

```
classification_models/
classification_block.py