gangweiX / ACVNet

[TPAMI 2024 & CVPR 2022] Attention Concatenation Volume for Accurate and Efficient Stereo Matching
MIT License
458 stars 63 forks source link

Dataset preparation #9

Open YuhsiHu opened 2 years ago

YuhsiHu commented 2 years ago

Thank you for your great work! There are many subsets of scene flow and KITTI dataset. Which should I download and how should I organize them? There will be an error even if I set dataset and datapath:

FileNotFoundError: [Errno 2] No such file or directory: '/home/xgw/data/KITTI_2012/frames_finalpass/TRAIN/C/0297/left/0006.png'

Because in kitti_dataset_1215.py:

def __init__(self, datapath, list_filename, training):
        self.datapath_15 = '/home/xgw/data/KITTI_2015/'
        self.datapath_12 = '/home/xgw/data/KITTI_2012/'
        self.left_filenames, self.right_filenames, self.disp_filenames = self.load_path(list_filename)
        self.training = training
        if self.training:
            assert self.disp_filenames is not None

Could you please specify the steps about how to organize data to meet the requirement? Thank you for your time!

gangweiX commented 2 years ago

see filenames/sceneflow_train.txt

YuhsiHu commented 2 years ago

I have seen it, thank you. There are many subsets of scene flow and KITTI datasets and the structure of these files differ from the path in your txt. Could you please specify the set that we download and the expected file structure? Thank you for your time!

bhu619 commented 1 year ago

If you use the scene flow datasets, refer to the txt file under the filenames folder that contains the path to the dataset, as well as the KITTI dataset.

I have seen it, thank you. There are many subsets of scene flow and KITTI datasets and the structure of these files differ from the path in your txt. Could you please specify the set that we download and the expected file structure? Thank you for your time!