Skip to content

Latest commit

 

History

History
23 lines (18 loc) · 860 Bytes

dataset.md

File metadata and controls

23 lines (18 loc) · 860 Bytes

Dataset Preparation

Preparing CO3Dv1 Dataset

Follow the directions to download and extract CO3Dv1 from here.

You will need to pre-process that dataset to extract the bounding boxes and camera poses. The bounding boxes are processed separately because it takes the most amount of time and should only be run once.

python -m preprocess.preprocess_co3dv1 --category all --precompute_bbox \
    --co3d_v1_dir /path/to/co3d_v1
python -m preprocess.preprocess_co3dv1 --category all \
    --co3d_v1_dir /path/to/co3d_v1

Preparing Your Own Dataset

For inference on your own video, you can use the CustomDataset class in relpose/dataset/custom.py. You will need to provide a directory of images and a directory of masks or bounding boxes. The masks are simply used to compute bounding boxes.