CodeMirror/waifu2x
1
0
Fork 0
mirror of synced 2024-06-01 10:39:30 +12:00
Code Issues Releases Wiki Activity
waifu2x/train.sh

14 lines
1.2 KiB
Bash
Raw Normal View History

update training script
2015-05-17 17:42:53 +12:00
#!/bin/sh
sync from internal repo - Memory compression by snappy (lua-csnappy) - Use RGB-wise Weighted MSE(R*0.299, G*0.587, B*0.114) instead of MSE - Aggressive cropping for edge region and some change.
2015-10-26 13:23:52 +13:00
th convert_data.lua
th train.lua -color rgb -random_half 1 -jpeg_sampling_factors 444 -color_noise 0 -overlay 0 -epoch 200 -method noise -noise_level 1 -crop_size 46 -batch_size 8 -model_dir models/anime_style_art_rgb -test images/miku_noisy.jpg -validation_ratio 0.1 -active_cropping_rate 0.5 -active_cropping_tries 10 -validation_crops 80
add support for RGB color space reconstruction - add new RGB model (models/anime_style_art_rgb). - RGB model can reduce color noise. - waifu2x uses this RGB model by default. You can use Y model with: $ th waifu2x.lua -model_dir models/anime_style_art -i input.png -o output.png $ th train.lua -color y ...
2015-06-23 05:27:28 +12:00
th cleanup_model.lua -model models/anime_style_art_rgb/noise1_model.t7 -oformat ascii
update training script
2015-05-17 17:42:53 +12:00
sync from internal repo - Memory compression by snappy (lua-csnappy) - Use RGB-wise Weighted MSE(R*0.299, G*0.587, B*0.114) instead of MSE - Aggressive cropping for edge region and some change.
2015-10-26 13:23:52 +13:00
th train.lua -color rgb -random_half 1 -jpeg_sampling_factors 444 -color_noise 0 -overlay 0 -epoch 200 -method noise -noise_level 2 -crop_size 46 -batch_size 8 -model_dir models/anime_style_art_rgb -test images/miku_noisy.jpg -validation_ratio 0.1 -active_cropping_rate 0.5 -active_cropping_tries 10 -validation_crops 80
add support for RGB color space reconstruction - add new RGB model (models/anime_style_art_rgb). - RGB model can reduce color noise. - waifu2x uses this RGB model by default. You can use Y model with: $ th waifu2x.lua -model_dir models/anime_style_art -i input.png -o output.png $ th train.lua -color y ...
2015-06-23 05:27:28 +12:00
th cleanup_model.lua -model models/anime_style_art_rgb/noise2_model.t7 -oformat ascii
update training script
2015-05-17 17:42:53 +12:00
sync from internal repo - Memory compression by snappy (lua-csnappy) - Use RGB-wise Weighted MSE(R*0.299, G*0.587, B*0.114) instead of MSE - Aggressive cropping for edge region and some change.
2015-10-26 13:23:52 +13:00
th train.lua -color rgb -random_half 1 -jpeg_sampling_factors 444 -color_noise 0 -overlay 0 -epoch 200 -method scale -crop_size 46 -batch_size 8 -model_dir models/anime_style_art_rgb -test images/miku_small_noisy.jpg -active_cropping_rate 0.5 -active_cropping_tries 10 -validation_ratio 0.1 -validation_crops 80
add support for RGB color space reconstruction - add new RGB model (models/anime_style_art_rgb). - RGB model can reduce color noise. - waifu2x uses this RGB model by default. You can use Y model with: $ th waifu2x.lua -model_dir models/anime_style_art -i input.png -o output.png $ th train.lua -color y ...
2015-06-23 05:27:28 +12:00
th cleanup_model.lua -model models/anime_style_art_rgb/scale2.0x_model.t7 -oformat ascii
sync from internal repo - Memory compression by snappy (lua-csnappy) - Use RGB-wise Weighted MSE(R*0.299, G*0.587, B*0.114) instead of MSE - Aggressive cropping for edge region and some change.
2015-10-26 13:23:52 +13:00
Reference in a new issue Copy permalink