Questins about json files

[emrekeles-arch]

Hello,

First of all, thank you for this excellent work. I have a few questions for you:

1)What does the "depth" in your dataset's JSON file represent?
2)How did you determine the "image_id" in the annotations section?

Thank you.

[Tajamul21]

Hello,

Thank you for your feedback!

1. The "depth" in the JSON file represents the no of channels.

2. The "image_id" is a unique identifier used to link annotations to their corresponding images. It is done usnig coco annotation tool

Best!

[emrekeles-arch]

@Tajamul21
Have you ever tried training using the csv file insted of json file ?
Is this possible with these codes and which parts of the code should I change ?

Thank you.

[Tajamul21]

The current code is specifically designed to work with COCO format datasets, which is why it relies on JSON files for annotations. If you want to use a CSV file instead, you'll need to write a new dataloader that can parse the CSV and convert it into the format expected by the rest of the code.

This would involve making changes primarily in the dataloader and potentially some other parts of the code where dataset handling occurs. If you're familiar with how COCO datasets are structured, you could map the information from the CSV into a similar structure for compatibility.