search

utiasASRL / deep_learned_visual_features

Deep learned features for long-term localization in Visual Teach and Repeat
BSD 3-Clause "New" or "Revised" License
54 stars 8 forks source link

key-points extract tool #1

Open zhengshunkai opened 2 years ago

zhengshunkai commented 2 years ago

Hi,Thanks for your great job. 1.Can you share a key-points extract and match tool by the great pth-model? So we can test this keypoints quickly.. 2.Can the model detect the key-points of Monocular image？

mona-robotics commented 1 year ago

@zhengshunkai Apologies for the late reply. I had failed to pay attention to there being issues posted. Thank you for your interest in the work.

1) I have posted a script that does simple feature extraction. Matching can be done by any method you prefer for matching descriptor vectors. https://github.com/utiasASRL/deep_learned_visual_features/blob/main/extract_features.py 2) Yes the network takes a single monocular image. I used stereo when training because I needed the depth of keypoints.