/unseenpose_correspondence_labelling

Primary LanguagePython

unseen_6d_pose_correpsondence_labelling

Dependencies

graspnetAPI

open3d==0.12.0.0

tensorboard

tqdm

torch

Minkowski Engine

Prepare codes

replace the

GRASPNET_ROOT = '/place/your/directory/to/synthetic/graspnet/dataset/here'
GRASPNET_ROOT_REAL = '/place/your/directory/to/real/graspnet/dataset/here'
LABEL_DIR = '/place/your/directory/to/label/synthetic/here'
LABEL_DIR_REAL = '/place/your/to/label/synthetic/here'

by the data set root and the directory you want to place the labels

GRASPNET_ROOT is the root directory to the synthetic dataset donated by https://graspnet.net/unseenpose.html GRASPNET_ROOT_REAL is the root directory to the real data donated by https://graspnet.net/datasets.html

Run the code with

bash ./generate_labels.sh $current_process_id $total_process_count $isreal

You can run multi processes simultaneously by having $total_process_count > 1. The total dataset will be divided into $total_process_count part.

$current_process_id means this process is processing the $current_process_idth part.

isreal means is True when your dataset real elsewise False