The datasets of gen_data.py

[pxiangwu]

The datasets of gen_data.py

[pxiangwu]

Hello,

Yes, you need to download all the 10 parts to obtain a complete dataset. In total there are 850 scenes.

[zhouwy-lv]

Thanks for your replay! In addition, the v1.0_mini dataset seems to have problems when running train_single_seq.
First of all, shuffle must be false to be executed.
trainloader = torch.utils.data.DataLoader(trainset, batch_size=BATCH_SIZE, shuffle=True, num_workers=num_workers)
But, The training datasize is 0.
While in train_muti_seq, shuffle can be true, and the datasize is 138.