How to generate training/test data for dissimilarity module?

[yxwang7]

Hi! I'm working on the pixel-wise anomaly detection and quite interested in your project! Now I'm trying to integrate other segmentation models in the SynBoost pipeline. However, I'm a little bit of confused about the data process scripts you've provided. In detail,

• is there any usage example of the three scripts provided under the data_preparation directory?
• In addition, I've found that the labels_with_ROI is provided for Fishyscapes as test set. Would you mind suggesting on how to generate these labels?

BTW, thank you for your kind notice and patience!