search

szagoruyko / cvpr15deepcompare

Code and models for "Learning to Compare Image Patches via Convolutional Neural Networks"
http://imagine.enpc.fr/~zagoruys/deepcompare.html
Other
468 stars 169 forks source link

Ask for the detailed information about the dataset yosemite? #34

Open yunyundong opened 6 years ago

yunyundong commented 6 years ago

def read_matches_file(filename): print(filename) data = np.loadtxt(filename, dtype=np.uint64) mask = data[:,1] == data[:,4] pairs = data[:,(0, 3)] return pairs[mask], pairs[np.logical_not(mask)]

for example, the txt file named ,m50_2000_2000_0.txt. what's the meaning of each column in the txt file? @szagoruyko how do we find the detailed describtion about the training data format.

yunyundong commented 6 years ago

I found each image, whose format is bmp, for example pathces0000.bmp in the yosemite folder, has 16*16 patches? how it links to the txt file? @szagoruyko

szagoruyko commented 6 years ago

@yunyundong see https://github.com/szagoruyko/cvpr15deepcompare/issues/20