Question about the used dataset

[wmj183363206]

Can the ShadowFormer process the dataset without gt_mask ?

[GuoLanqing]

You can use an existing shadow detector to pre-predicted a shadow mask, then input it into the ShadowFormer. Or you can re-train the model without the shadow mask.

[wmj183363206]

OK. Thank you for your reply.

[wmj183363206]

你好，我想问一下如果我是用宽度为960高度为720的图片，这个训练指令可以是这样的吗？ python train.py --warmup --win_size 12 --train_ps 720

[GuoLanqing]

The train_ps should be a multiple of the win_size * 8. Thus you can train with the win_size=10. Alternatively, the train_ps=720 is very large, leading to a large GPU memory. You can choose a smaller train_ps.

[wmj183363206]

Thank you for your patient reply. ^ . ^

[hnrna]

The train_ps should be a multiple of the win_size * 8. Thus you can train with the win_size=10. Alternatively, the train_ps=720 is very large, leading to a large GPU memory. You can choose a smaller train_ps.

Have you experimented with the effect of adjusting the parameters --win_size and --train_ps on the effect of the model?

In addition does the program limit the upper limit of the parameters --win_size and --train_ps?