Skip to content

Latest commit

 

History

History
51 lines (44 loc) · 1.48 KB

prepare_dataset.md

File metadata and controls

51 lines (44 loc) · 1.48 KB

Dataset Prepare

Pre-trained CLIP Models

Download the pre-trained CLIP models (VIT-B-16.pt) and save them to the pretrained folder.

Pascal VOC 2012

  • First download the Pascal VOC 2012 datasets use the scripts in the data dir.
cd data
sh download_and_convert_voc12.sh
  • Then download SBD annotations from here.

The folder structure is assumed to be:

WeakCLIP
├── data
│   ├── download_and_convert_voc12.sh
│   ├── VOCdevkit
│   │   ├── VOC2012
│   │   │   ├── JPEGImages
│   │   │   ├── SegmentationClass
│   │   │   ├── SegmentationClassAug
│   │   │   ├── SegmentationClassAugPseudoMaskMCT

COCO 2014

  • First download the COCO 2014 datasets use the scripts in the data dir.
cd data
sh download_and_convert_coco.sh
cp val_5000.txt COCO14/voc_format

The folder structure is assumed to be:

WeakCLIP
├── data
│   ├── download_and_convert_coco.sh
│   ├── VOCdevkit
│   ├── COCO14
│   │   ├── images
│   │   ├── voc_format
│   │   │   ├── class_labels
│   │   │   ├── train.txt
│   │   │   ├── val.txt
│   │   │   ├── val_5000.txt
│   │   │   ├── cocoPGTMCT