How does the raw dataset folder organize?

[bridgeqiqi]

Thank you for your great work which inspires me a lot. The dataset preprocessing code seems to rely on the folder organization. Could you provide more details (e.g. the number of raw images and processed images)?

[dylran]

We used the original folder organization in the public datasets. For example, in line 85 of preprocess_shtech.py, we provide the image directory. Then in line 106, we derive the path to the ground truth from the respective image path. Since different datasets have different folder organization structures, you must use the appropriate preprocessing code.