Skip to content

Request for fine-tuning guide on my custom dataset (coco format) #8

@MHkimmm

Description

@MHkimmm

Hi, first of all, thank you for releasing this great project!
I’m trying to fine-tune the model on my own dataset (custom images and annotations), but I couldn’t find a clear guide or example for doing so.

Could you please provide a short tutorial or documentation on how to train/fine-tune this model on a custom dataset (of COCO format)?
Even a minimal example or reference config file would be very helpful.

What I've Tried

  • I successfully set up the environment and ran inference on the provided pretrained model.

  • I’m familiar with the Detectron2 training pipeline (e.g., using register_coco_instances, DefaultTrainer etc.), but since your repo seems to have some custom training modules, configs and strategies, I’m unsure how to properly integrate my dataset into this workflow.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions