naver/r2d2

Issues

• Confusion about the different architectures model size

  #54 opened 4 months ago by ShowStopperTheSecond
  1

• How to evaluate the model on Hpatches and visualize the results using .ipynb?

  #22 opened 4 years ago by whubaichuan
  5

• How to generate .npy file ?

  #29 opened 4 years ago by water-wave
  3

• If I want to train, where to write the dataset's path?

  #53 opened 5 months ago by Ma0zii
  0

• How to understand AP loss in this paper? What's the meaning of quantizer?

  #43 opened 3 years ago by reacher-l
  1

• how to use the file .r2d2?

  #51 opened a year ago by ErniuC
  0

• Pipeline for Optical Flow

  #50 opened a year ago by kkaytekin
  0

• Reliability loss converges to 0.5 instead of 0.

  #16 opened 4 years ago by MarviB16
  2

• Training model with only 1 dataset

  #48 opened 2 years ago by dfvr1994
  0

• Require for the image data of Zero

  #47 opened 2 years ago by Katherinaxxx
  2

• repeatability on the Oxford dataset.

  #45 opened 3 years ago by reacher-l
  1

• Codes for making datasets

  #1 opened 5 years ago by ChunhuanLin
  2

• Training on custom datasets

  #46 opened 3 years ago by ericzzj1989
  0

• Could you release the calculation of matching performance ? Such as Repeatability and M-Scores

  #8 opened 4 years ago by GabbySuwichaya
  5

• About matching score

  #44 opened 3 years ago by reacher-l
  2

• Testing

  #11 opened 4 years ago by Zumbalamambo
  8

• Matching Score

  #42 opened 3 years ago by jyy-1998
  0

• Is there a way to change descriptor size?

  #41 opened 3 years ago by UmutCosk
  0

• Does the training code support multi-gpu training?

  #37 opened 4 years ago by pixelsmaker
  1

• when is the loss_from_ap in the ReliabilityLoss class called ?

  #35 opened 4 years ago by hichem-abdellali
  7

• The problem downloading the revisitop1m dataset

  #36 opened 4 years ago by pixelsmaker
  4

• Some details about Oxford dataset

  #39 opened 4 years ago by ThompsonHe
  1

• Why the time of feature extraction is more than superpoint?

  #40 opened 4 years ago by OldAAAA
  1

• Matching-score

  #38 opened 4 years ago by pixelsmaker
  1

• Learn from Correspondences instead of aflow

  #26 opened 4 years ago by ShrutheeshIR
  4

• Explanation of this implementation

  #34 opened 4 years ago by Guptajakala
  0

• question about two losses

  #33 opened 4 years ago by Jemmagu
  1

• Welcome to the Learning Group!!!

  #23 opened 4 years ago by whubaichuan
  3

• Some questions of sampler.py

  #32 opened 4 years ago by UncleChuanchuan
  1

• What if I want it work in a low resolution situation?

  #28 opened 4 years ago by lvgeng
  1

• Is the model structure in the code the same as described in the paper？

  #31 opened 4 years ago by kkzh2313
  2

• How to extract patch descriptor only?

  #30 opened 4 years ago by ducha-aiki
  1

• Getting NaN at reliability loss occasionally during training

  #21 opened 4 years ago by GabbySuwichaya
  8

• Visualize heatmap

  #19 opened 4 years ago by dzungdoan6
  1

• Unable to extract features with pretrained models under conda

  #20 opened 4 years ago by iSecret2048
  1

• code trouble

  #27 opened 4 years ago by water-wave
  1

• Generate image pairs using random homography

  #17 opened 4 years ago by hanyunlong233
  2

• Could you tell me your training and feature extracting setting for local feature challenge in visual localization benchmark

  #14 opened 4 years ago by hongsiyu
  2

• Doubts about reliability performance in custom image

  #18 opened 4 years ago by ShrutheeshIR
  4

• Typo in reliability loss equation in paper

  #15 opened 4 years ago by rsokl
  1

• How to solve warning about "torch.nn.functional.grid_sample()" due to Pytorch version 1.3.0.

  #13 opened 4 years ago by GabbySuwichaya
  2

• Questions about the time cost of R2D2 feature extraction

  #9 opened 4 years ago by Jemmagu
  6

• Images used for style transfer

  #12 opened 4 years ago by uzpaka
  4

• About the difference between MMA and M-Score

  #10 opened 4 years ago by zehongs
  0

• How can one pixel of the image correspond to a feature point and a descriptor?

  #6 opened 5 years ago by ahong007007
  1

• How to reproduce the detector repeatability on the Oxford dataset ?

  #7 opened 5 years ago by djkim3
  1

• Can't access Dataset address

  #5 opened 5 years ago by djkim3
  1

• Reproduce the results in the paper

  #4 opened 5 years ago by pengsida
  2

• How do you implement feature extraction for large-sized images?

  #3 opened 5 years ago by ahong007007
  1

• Looking forward to more example programs, thanks again.

  #2 opened 5 years ago by ahong007007
  1