From bf22450b8d37a24433d07dad5a973f0a80a7dd34 Mon Sep 17 00:00:00 2001 From: "Alex J. Champandard" Date: Sat, 29 Oct 2016 14:15:24 +0200 Subject: [PATCH] Update documentation for new --train usage, minor improvements. --- README.rst | 18 +++++++++--------- ...own_example.gif => StreetView_example.gif} | Bin 2 files changed, 9 insertions(+), 9 deletions(-) rename docs/{Chinatown_example.gif => StreetView_example.gif} (100%) diff --git a/README.rst b/README.rst index c025245..79358e4 100644 --- a/README.rst +++ b/README.rst @@ -58,13 +58,13 @@ Pre-trained models are provided in the GitHub releases. Training your own is a rm -f ne4x*.pkl.bz2 # Pre-train the model using perceptual loss from paper [1] below. - python3.4 enhance.py --train --scales=2 --epochs=50 \ + python3.4 enhance.py --train "data/*.jpg" --scales=2 --epochs=50 \ --perceptual-layer=conv2_2 --smoothness-weight=1e7 --adversary-weight=0.0 \ --generator-blocks=4 --generator-filters=64 # Train the model using an adversarial setup based on [4] below. - python3.4 enhance.py --train --scales=2 --epochs=250 \ - --perceptual-layer=conv5_2 --smoothness-weight=2e4 --adversary-weight=1e3 \ + python3.4 enhance.py --train "data/*.jpg" --scales=2 --epochs=250 \ + --perceptual-layer=conv5_2 --smoothness-weight=2e4 --adversary-weight=2e5 \ --generator-start=5 --discriminator-start=0 --adversarial-start=5 \ --discriminator-size=64 @@ -119,11 +119,11 @@ This code uses a combination of techniques from the following papers, as well as 3. `Deeply-Recursive Convolutional Network for Image Super-Resolution `_ 4. `Photo-Realistic Super-Resolution Using a Generative Adversarial Network `_ -Special thanks for their help in various ways: +Special thanks for their help and support in various ways: -* Eder Santana — Support and encouragement, feedback and ideas on `sub-pixel deconvolution `_. -* Andrew Brock — This sub-pixel layer code is based on `his project repository `_. -* Casper Kaae Sønderby — For suggesting a more stable alternative to sigmoid + log. +* Eder Santana — Discussions, encouragement, and his ideas on `sub-pixel deconvolution `_. +* Andrew Brock — This sub-pixel layer code is based on `his project repository `_ using Lasagne. +* Casper Kaae Sønderby — For suggesting a more stable alternative to sigmoid + log as GAN loss functions. 4. Troubleshooting Problems @@ -160,9 +160,9 @@ It seems your terminal is misconfigured and not compatible with the way Python t **FIX:** ``export LC_ALL=en_US.UTF-8`` -.. image:: docs/Chinatown_example.gif +.. image:: docs/StreetView_example.gif -**Example #4** — China Town: `view comparison `_ in 24-bit HD, `original photo `_ CC-BY-SA @cyalex. +**Example #4** — Street View: `view comparison `_ in 24-bit HD, `original photo `_ CC-BY-SA @cyalex. ---- diff --git a/docs/Chinatown_example.gif b/docs/StreetView_example.gif similarity index 100% rename from docs/Chinatown_example.gif rename to docs/StreetView_example.gif