dataset

API details.

---

WxBSDataset

 WxBSDataset (root_dir:str, subset:str='test', version='v1.1',
              download:bool=True)

Wide multiple baselines stereo dataset.

dset = WxBSDataset('.WXBS', download=True)
print (dset[10].keys())
print (dset[10]['name'])

dict_keys(['img1', 'img2', 'pts', 'errors', 'name'])
WGALBS/submarine2

---

EVDDataset

 EVDDataset (root_dir:str, subset:str='test', version='v1.0',
             download:bool=True)

Extreme View Dataset.

Our dataset returns tuples of: image1, image2, labelled GT correspondences, cross-validation errors

dset = EVDDataset('.EVD', download=True)
print (dset[0].keys())
print (dset[0]['name'])
print (dset[0]['H'])

Downloading http://cmp.felk.cvut.cz/wbs/datasets/EVD.zip to .EVD2/EVD.zip
# Extracting data .EVD2

dict_keys(['img1', 'img2', 'img1_shape', 'img2_shape', 'H', 'name'])
adam
[[ 1.60197402e-01 -9.50942558e-02  2.72132086e+02]
 [ 1.13716755e-02  8.83741135e-01  4.04505014e+00]
 [-7.88552590e-05 -3.01914075e-05  1.00000000e+00]]

100%|█████████████████████████████████████████████████████████████| 28531000/28531000 [00:04<00:00, 6793575.68it/s]