GitHub1y
Learning to Generate Language-supervised and Open-vocabulary Scene Graph using Pre-trained Visual-Semantic Space
Environment. This repo requires Pytorch>=1.9 and torchvision. ├─coco │ ├─annotations │ ├─train2017 │ ├─val2017 │ └─zero-shot └─VG150 ├─VG_100K ├─weak_supervisions ├─image_data.json ├─VG-SGG-dicts-with ...
GitHub7mon
data.md
vg_stats.pt # frequency bias ├─ zeroshot_triplet.pytorch ├─ stanford_filtered/ ├─ image_data.json ├─ VG-SGG-dicts.json # add split_GLIPunseen ├─ VG-SGG.h5 download vg150 following ...