Code processing data

[ltphuongunited]

Can you public source code to process original video data of UVG? Thank you so much.

[boheumd]

Hi, the data processing uses the image compression technique to compress the UVG keyframes. It consists of the following steps:

1. extract frames from original YUV videos on the UVG dataset, and crop 1080x1920 frames to 1024x1920
2. apply the image compression method on the selected keyframes with gop_size = 8. You can follow this repo or this repo for the image compression method.
3. split the decoded (compressed) key-frames into 256x320 patches, which will be regarded as the input frames for the D-NeRV

Sorry, the original source code of data processing is not cleaned yet and is still very messy, it will take some time to clean it.

[hahazh]

hi, which kinds of crops do you perform,regarding to "crop 1080x1920 frames to 1024x1920" , did you adopt center crop?

[boheumd]

Hi, I adopt the upper left crop instead of the center crop.