CAI, Rizhao

如题。请问在训练过程中，batch_size一般为1吗？因为不同的样本似乎有不同的动作选择次数，一次性多个batch训练是不可行的，所以batch_size一般都是1，对吗？

In the paper "Deep Spatial Gradient and Temporal Depth Learning for Face Anti-spoofing", it is said that a double-modal dataset (DMAD) is collected. Is this publicly available?

If I want to prune my model with object detection models, what should I pay attention to? What should I change?

Question about the loss calculation

2

Hi, Thanks for contributing the source code. I have some questions about the loss calculation in https://github.com/zhiqiangdon/online-augment/blob/9d46ab4789f6cd91b84fd05445866f3b3927e94f/train_aug_stn.py#L154  According to the paper, for training the augmentation network, the adversarial losses...

which dataset is the pretrained model pretrained from?

1

which dataset is the pretrained model pretrained from? Also, is your code the implementation of any paper? If yes, what paper? Thanks for your sharing!

The code does not accommodate with the latest tegrastats

1

print("没有符合条件的输出-2") reg2 = re.compile( r'RAM (\d+).+?CPU \[(.+?)\] EMC_FREQ (\d+)%@.+?GR3D_FREQ (\d+)% *') However, now the tegrastat will also output the information of VDD and so on like this RAM 2020/7852MB (lfb...

Can you reproduce the results as shown in the paper?

1

Can you reproduce the results as shown in the paper? Or very closed results(e.g.

where to download the UVAD dataset?