search

blakechen97 / SASA

SASA: Semantics-Augmented Set Abstraction for Point-based 3D Object Detection
Apache License 2.0
90 stars 14 forks source link

something about k-fold cv #5

Open OuyangJunyuan opened 2 years ago

OuyangJunyuan commented 2 years ago

as mention in this work:

For the submission to KITTI test server, we
follow the training protocol mentioned in (Shi et al. 2020a),
where 80% labeled point cloud images are used for training
and the rest 20% images are used for validation.

it means you use 80% of that to train and 20% to turn parameters, and then summited the best one to KITTI test server? and where could I get this 5-fold cv split files?( or just done by simply skip 4 sample,if so i can do by myself)

blakechen97 commented 2 years ago

Hi,

Yes, we randomly pick 80% labeled examples for training and use the remaining 20% for validation. You can simply write a script to generate the split or just follow the strategy you've mentioned.

OuyangJunyuan commented 2 years ago

Hi,

Yes, we randomly pick 80% labeled examples for training and use the remaining 20% for validation. You can simply write a script to generate the split or just follow the strategy you've mentioned.

awesome! thanks for rapid reply