vismatch.im_models.handcrafted

Classes

HandcraftedBaseMatcher([device])

This class is the parent for all methods that use a handcrafted detector/descriptor, It implements the forward which is the same regardless of the feature extractor of choice.

OrbNNMatcher([device, max_num_keypoints, ...])

SiftNNMatcher([device, max_num_keypoints, ...])

class vismatch.im_models.handcrafted.HandcraftedBaseMatcher(device='cpu', **kwargs)[source][source]

Bases: BaseMatcher

This class is the parent for all methods that use a handcrafted detector/descriptor, It implements the forward which is the same regardless of the feature extractor of choice. Therefore this class should NOT be instatiated, as it needs its children to define the extractor/detector.

Parameters:

device (str)

static preprocess(im_tensor)[source][source]
Parameters:

im_tensor (Tensor)

Return type:

ndarray

class vismatch.im_models.handcrafted.SiftNNMatcher(device='cpu', max_num_keypoints=2048, lowe_thresh=0.75, *args, **kwargs)[source][source]

Bases: HandcraftedBaseMatcher

Parameters:

device (str)

class vismatch.im_models.handcrafted.OrbNNMatcher(device='cpu', max_num_keypoints=2048, lowe_thresh=0.75, *args, **kwargs)[source][source]

Bases: HandcraftedBaseMatcher

Parameters:

device (str)