I have a small problem about semantic segmentation part, I found S3DIS training data is in the form of hdf5 file, but in the testing phase, S3DIS data is stored as .npy file and transfered to .txt file for test, I wonder is there any reason to use different data format in traing and testing phase separately?
I have a small problem about semantic segmentation part, I found S3DIS training data is in the form of hdf5 file, but in the testing phase, S3DIS data is stored as .npy file and transfered to .txt file for test, I wonder is there any reason to use different data format in traing and testing phase separately?