search

berkeley-hipie / HIPIE

[NeurIPS2023] Code release for "Hierarchical Open-vocabulary Universal Image Segmentation"
https://people.eecs.berkeley.edu/~xdwang/projects/HIPIE/
MIT License
271 stars 20 forks source link

datasets #14

Open Richardg706 opened 8 months ago

Richardg706 commented 8 months ago

i really really appreciate your wonderful great job very much!! but i am so confused with the odise implement to get the needed datasets such as coco panoptic and so on. could you public the datasets that must produced by odise which i really couldn't handle? much thanks ,appreciate !!

KKallidromitis commented 7 months ago

Hi for training and evaluation we used public datasets which can be found here:

RefCOCO - Link Pascal Parts - Link PACO - Link Object365 - Link COCO Panoptic - Link SegInW - Link ODinW - Link

If you need help setting up coco panoptic I include a link which contains a zip with files for: coco_2017_train_panoptic_with_sem_seg / coco_2017_val_panoptic_with_sem_seg

These include: coco/panoptic_train2017 coco/panoptic_semseg_train2017 coco/panoptic_val2017 coco/panoptic_semseg_val2017 coco/annotations/panoptic_train2017.json coco/annotations/panoptic_val2017.json

Richardg706 commented 6 months ago

Thank you very much !!!