search

mihaidusmanu / d2-net

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features
Other
764 stars 163 forks source link

The different between SoftDetectionModule and HardDetectionModule #31

Closed ahong007007 closed 4 years ago

ahong007007 commented 4 years ago

The training phase in the open source code uses the SoftDetectionModule in model.py. The test uses the HardDetectionModule of model_test.py. What is the difference? Thanks.

mihaidusmanu commented 4 years ago

See section 3.2 of the paper - https://arxiv.org/pdf/1905.03561.pdf, paragraphs "Hard Feature Detection" and "Soft Feature Detection" respectively.

TL;DR: the soft feature detection is a soft translation of the hard feature detection using convolutional soft-max; the soft formulation is only used at train time.