Dear Authors,
Thank you for your remarkable work. I found your paper highly insightful and very valuable to my research.
I had a few questions and would greatly appreciate your clarification:
First, regarding dataset construction, did you annotate attributes for the entire DSEC dataset or only for a subset?
Second, in the appendix, you mention that “the dataset is partitioned into a training split of 4,433 scenes and a test split of 1,134 scenes.” I would like to better understand how these “scenes” map to the original DSEC data, since the total amount of data across the files appears larger than those numbers.
I also wanted to ask whether you are planning to release the dataset construction code, the checkpoint of your model, and the models you re-implemented under the same DETR transformer and grounding head framework.
Best regards,
Mohamad Alansari
Dear Authors,
Thank you for your remarkable work. I found your paper highly insightful and very valuable to my research.
I had a few questions and would greatly appreciate your clarification:
First, regarding dataset construction, did you annotate attributes for the entire DSEC dataset or only for a subset?
Second, in the appendix, you mention that “the dataset is partitioned into a training split of 4,433 scenes and a test split of 1,134 scenes.” I would like to better understand how these “scenes” map to the original DSEC data, since the total amount of data across the files appears larger than those numbers.
I also wanted to ask whether you are planning to release the dataset construction code, the checkpoint of your model, and the models you re-implemented under the same DETR transformer and grounding head framework.
Best regards,
Mohamad Alansari