Skip to content

No_of_crops for custom dataset for both training and inference. #52

Description

@Sanath1998

HI @Stephen0808, @Shuweis , @TianheWu ,

In KADID10k dataset, images are of size 512 x 384 (width x height). In the code no_of_crops given was 20 or this dataset.
What is the purpose of taking 20 crops, is there any particular reason. I could generalize my understanding stating that random 20 crops of size 224 x 224 covers almost 80% of image.

How many no_of_crops is suggested for 9k resolution image. Is the any particular formula for that or any other reasons?

This understanding is very much helpful for training activities.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions