zhu-xlab/SSL4EO-S12

Issues

• Documentation: Code Entry Point and Data Preprocessing Details for `B13_rn18_moco_0099_ckpt.pth`

  #28 opened 3 months ago by douglasmacdonald
  4

• Pre-trained weights for fine-tuned datasets

  #27 opened 6 months ago by soc12
  1

• misalignment between SSL4EO-S12 and real world maps

  #24 opened 8 months ago by wangyi111
  1

• Question about linear BE DINO

  #26 opened 7 months ago by jdilger
  2

• Question regarding locations between S1/S2

  #25 opened 8 months ago by sunny1401
  2

• Sentinel-1 and Sentinel-2 image alignment for Self-Supervised Segmentation pre-training

  #22 opened 9 months ago by KIselborn
  3

• Micro mAP calculation

  #23 opened 10 months ago by Ehzoahis
  4

• Potential loss in accuracy from using the compressed 8-bit dataset

  #18 opened a year ago by RyanAmaudruz
  7

• Questions around data preparation and hyperparameters for linear probing

  #21 opened a year ago by RyanAmaudruz
  2

• Dataloader Issue

  #20 opened a year ago by moonboy12138
  4

• Reproducing the segmentation results (DFC2020)

  #19 opened a year ago by RyanAmaudruz
  2

• Normalization for S1?

  #17 opened a year ago by bengmstrong
  4

• Difficulties Unzipping Downloads

  #16 opened a year ago by tomburgert
  2

• General Query regarding paper

  #15 opened a year ago by vishalned
  2

• Reproducing the results of linear probing

  #14 opened a year ago by Ehzoahis
  5

• Clarification regarding loading pretrained models

  #13 opened a year ago by pcvinayak
  1

• could you please public the code of fine-tune.py from mae and the training log？

  #11 opened a year ago by DDKdakuan
  1

• SAR preprocessing

  #12 opened a year ago by edornd
  2

• Change Detection using MoCo SSL for S1 SAR 2 bands

  #7 opened a year ago by vaasuSS
  5

• What is the RGB version?

  #6 opened a year ago by Zhou2019
  10

• Spatial bias in dataset

  #3 opened a year ago by calebrob6
  2

• Training loss keeps increasing when I removed the distribute computation code and trained with Single GPU

  #8 opened a year ago by ftongss
  3

• DINO finetune

  #9 opened a year ago by wgd4510
  2

• The number of channels in bigearthnet (L2a) and ssl4eo (L1c) is different.

  #10 opened a year ago by DDKdakuan
  0

• Can't pickle local object 'make_lmdb.<locals>.<lambda>'

  #4 opened 2 years ago by vaasuSS
  5

• Data normalization for pretrained models

  #2 opened 2 years ago by clkruse
  4

• License for pretrained models

  #1 opened 2 years ago by adamjstewart
  3