search

llmpass / RSTT

Official pytorch implementation of paper "RSTT: Real-time Spatial Temporal Transformer for Space-Time Video Super-Resolution"
117 stars 20 forks source link

about train config #8

Closed Winnie202 closed 1 year ago

Winnie202 commented 1 year ago

Excuse me, I am training my own super resolution data. The HR_image size is 720 * 1080. What relevant parameter settings need to be changed?crop_size need to change the size to 720?

2022-12-14 10-32-41屏幕截图

llmpass commented 1 year ago

Hi, your setting is correct. Probably, you can keep the crop size as 256. Please see the example at https://github.com/llmpass/RSTT/blob/927a1cf32af1469c77395aca59b12db0bdc9ff6c/data/Vimeo.py.

Winnie202 commented 1 year ago

Hi, your setting is correct. Probably, you can keep the crop size as 256.

Thank you for your reply. In addition, the size of my training data image format is 8GB. After the "create_lmdb.py" is converted to lmdb, but the size is close to 80GB. Is this normal?

llmpass commented 1 year ago

Hi, your setting is correct. Probably, you can keep the crop size as 256.

Thank you for your reply. In addition, the size of my training data image format is 8GB. After the "create_lmdb.py" is converted to lmdb, but the size is close to 80GB. Is this normal?

Please don't just use the file size to judge if the process is correct or not. I think a better way is to write a piece of code, for example a notebook to check if the entries in this lmdb is what you expected or not.

Winnie202 commented 1 year ago

Hi, your setting is correct. Probably, you can keep the crop size as 256.

Thank you for your reply. In addition, the size of my training data image format is 8GB. After the "create_lmdb.py" is converted to lmdb, but the size is close to 80GB. Is this normal?

Please don't just use the file size to judge if the process is correct or not. I think a better way is to write a piece of code, for example a notebook to check if the entries in this lmdb is what you expected or not.

Yes, I have verified that the content in the lmdb is correct, but the file becomes too large after being converted to the lmdb format, and my computer does not have enough space. I wonder if this is normal

llmpass commented 1 year ago

Hi, your setting is correct. Probably, you can keep the crop size as 256.

Thank you for your reply. In addition, the size of my training data image format is 8GB. After the "create_lmdb.py" is converted to lmdb, but the size is close to 80GB. Is this normal?

Please don't just use the file size to judge if the process is correct or not. I think a better way is to write a piece of code, for example a notebook to check if the entries in this lmdb is what you expected or not.

Yes, I have verified that the content in the lmdb is correct, but the file becomes too large after being converted to the lmdb format, and my computer does not have enough space. I wonder if this is normal

If this is the case, probably you can reduce the number of entries.