Official implementation of our TIV'23 paper: Occupancy-MAE: Self-supervised Pre-training Large-scale LiDAR Point Clouds with Masked Occupancy Autoencoders
Apache License 2.0
251
stars
18
forks
source link
Questions about how voxel-MAE can be combined with detection methods #5
Hello, after reading your paper, it is an excellent job. There is a confusion about how the pre-trained method is combined with other detection methods and how the encoder's weights are used？
Hello, after reading your paper, it is an excellent job. There is a confusion about how the pre-trained method is combined with other detection methods and how the encoder's weights are used？