zhiyuanyou / SAFECount

[WACV 2023] Few-shot Object Counting with Similarity-Aware Feature Enhancement
Other
124 stars 22 forks source link

Hello, how to generate the train.json file for training your own dataset #7

Closed 1274739295 closed 1 year ago

1274739295 commented 1 year ago

We would like to train our own dataset after reproducing your paper. I wonder if you can provide us with the train.json of how to generate our own dataset, and if you can tell us how to get the mask.png file in the mall dataset?

zhiyuanyou commented 1 year ago
  1. There are two kinds of train.json:
    • Annotated with boxes. In this case, you should annotate your own data with boxes circled every object (e.g., train.json).
    • Annotated with points. In this case, you should annotate your own data with points in every object center (e.g., train.json).

After annotation, you could generate a json file just following the structure of train.json.

  1. The mask.png is under the Mall dataset, (i.e., perspective_roi.mat). I did some transformation to transform it to a png image.