How to pre-process the test split to submit to the Scannet benchmark

facebookresearch / ContrastiveSceneContexts

Code for CVPR 2021 oral paper "Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts"

MIT License

221 stars 28 forks source link

How to pre-process the test split to submit to the Scannet benchmark #14

Closed KangchengLiu closed 3 years ago

KangchengLiu commented 3 years ago

Hi, for ScanNet detection data generation, in https://github.com/facebookresearch/votenet/tree/master/scannet, only the preprocessing for train/val data is provided (000-706). But the test data (707-806) also need to be processed for testing on the benchmark. Please provide relevant pre-processing for the test data (707-806). Thanks.

Sekunde commented 3 years ago

Hi,

We use the same script for generating the test data; you can download the processed test data from ScanNet Detection Test Data