About download 3DLoMatch

prs-eth / OverlapPredator

[CVPR 2021, Oral] PREDATOR: Registration of 3D Point Clouds with Low Overlap.

https://shengyuh.github.io/predator/index.html

MIT License

512 stars 73 forks source link

About download 3DLoMatch #44

Closed WuHaoYu21 closed 2 years ago

WuHaoYu21 commented 2 years ago

hi , i want to download 3DLoMatch dataset, but i don't know how to download and use it. The 3DLoMatch and 3DMatch are the same dataset? or some differences? I know these have different overlaps, but i want to use 3DLoMatch. Does 3DLoMatch and 3DMatch have same train sets, and 3DLoMatch has another test sets? Or i train a Model by 3DMatch, and test it by 3DLoMatch?

Looking forward to your reply, thank you!

ShengyuH commented 2 years ago

hi,

3DLoMatch is part of 3DMatch dataset, as you said, they have different overlap ratios.
you can simply threshold the overlap ratio and include pairs having 10-30% overlap ratios into training/validation.
From here you can find the 3DLoMatch benchmark.

Best, Shengyu

WuHaoYu21 commented 2 years ago

Thank you. So, 3DLoMatch is part of 3DMatch. If i want to use 3DLoMatch, i should threshold the overlap ratio and include pairs having 10-30% overlap ratios when i dataloader. Sorry, i also have some questions. In Predator, when test in 3DLoMatch , the train dataset is 3DMatch or 3DLoMatch? And i don't know what is the 'benchmarks', a dataset or something else? How to use the 'benchmarks'? Sorry, my English is not good, would you mind using Chinese? Or contact you by e-mail?

Looking forward to your reply, thank you very much!

ShengyuH commented 2 years ago

hi,

The training split used in Predator includes pairs having 10-30% overlap ratios
You can imagine benchmark as a competition, it includes the test dataset and evaluation metrics. Please refer to our evaluation part to see how we make use of the "benchmark".

Best, Shengyu

WuHaoYu21 commented 2 years ago

Thank you, Sorry, the training split includes (10%-30% and >30%) or (10%-30%), which one? If i want to using 3DLoMatch, first, i running 'sh scripts/download_data_weight.sh' to download 3DMatch preprocessed dataset, second, i training model by download dataset(threshold (10%-30% and >30%) or (10%-30%)), last, i evaluate this model by 'the benchmark'. am i right?

Thank you very much! Looking forward to your reply.

ShengyuH commented 2 years ago

the training split includes 10%-30% and >30%.

WuHaoYu21 commented 2 years ago

Does your preprocessed 3DMatch dataset includes 10%-30% and >30%? So when i training a model, i only need to using all training dataset (don't set the threshold, because the all training dataset includes 10%-30% and >30%).
And if i want to test a model , i only using the 'benchmarks' to test it . Does my understand is right?

ShengyuH commented 2 years ago

Yes. Feel free to reopen if needed.