Use coco_20k data?

[longzeyilang]

Hi, what should I trianing RoMa with coco_20k data augmentation?
like https://github.com/verlab/accelerated_features/blob/main/modules/dataset/augmentation.py
https://github.com/verlab/accelerated_features/blob/main/modules/dataset/augmentation.py
how should I revise?

[longzeyilang]

I do not want use megadepth and scannet data, just like coco_20k data?

[Parskatt]

Basically you need to update the dataloader and loss. I used to have coco in DKM training. Wouldnt really recommend it though.

[longzeyilang]

why?

[Parskatt]

Because if you train dense matchers on homography they only do well on homography.

[longzeyilang]

ok , I want to train my own dataset, about 128*128 image size

[longzeyilang]

According to you, I want use scannet data to crop 128*128 image size to train. and detection in my own dataset. is it correct?