Cleaning up. Updating License and Readme

README.md

@@ -4,7 +4,9 @@ Handwriting... with style!
 
 This is the code for the paper "Text and Style Conditioned GAN for the Generation of Offline-Handwriting Lines" published at BMVC 2020. https://arxiv.org/abs/2009.00678
 
-This was originally Brian Davis's summer 2019 internship project at Adobe.
+This was originally Brian Davis's summer 2019 internship project at Adobe (https://github.com/adobe-research/hw_with_style). It was then extended afterwards (while at BYU) and finally published.
+
+The trained models (snapshots) will be available as a file in the release.
 
 Code structure based on victoresque pytorch template.
 
@@ -29,7 +31,8 @@ Encoder: `python train.py -c configs/cf_IAM_auto_2tight_newCTC.json`
 
 Then the generator can be trained: `python train.py -c configs/cf_IAMslant_noMask_charSpecSingleAppend_GANMedMT_autoAEMoPrcp2tightNewCTCUseGen_balB_hCF0.75_sMG.json`
 
-RIMES will use
+
+The RIMES dataset will use:
 
 HWR: `python train.py -c configs/cf_RIMESLines_hwr_cnnOnly_batchnorm_aug.json`
 
@@ -79,7 +82,10 @@ Some modes need the datasets styles in a pickle. Use `get_styles.py` to extract
  │ ├── author_hw_dataset.py - This sorts instances by author and has a 'a_batch_size', which is how many by each author should be in the batch (batch_size is number of authors)
  │ ├── author_word_dataset.py - Same as above only at word level instead of line level
  │ ├── (the "mixed_author" datasets have different splits which mix authors over the splits)
- │ ├── font_dataset.py - dataset to render fonts as images. Required TextFlow (and is commented out in data_loaders.py)
+ │ ├── author_rimes_dataset.py - Same as author_hw_dataset.py, but for RIMES words
+ │ ├── author_rimeslines_dataset.py - Same as author_hw_dataset.py, but for RIMES lines
+ │ ├── synth_hw_dataset.py - For training a HWR model using the generated images
+ │ ├── create_synth_text_dataset.py - Generate images for synth_hw_dataset.py (although it can generate them on the fly)
  │ └── test*.py - This are scripts to run through a dataset and simply display what's being returned. For debugging purposes.
  │
  ├── logger/ - for training process logging
@@ -123,7 +129,7 @@ Some modes need the datasets styles in a pickle. Use `get_styles.py` to extract
  │
  ├── saved/ - default checkpoints folder
  │
- ├── configs/ - configuration files to reproducibility
+ ├── configs/ - configuration files for reproducibility
  ├── old_configs/ - all the other config files I've used during development/other projects
  │
  ├── trainer/ - trainers

base/base_data_loader.py

@@ -1,3 +1,4 @@
+#from https://github.com/victoresque/pytorch-template
 from copy import copy
 import numpy as np
 

base/base_model.py

@@ -1,3 +1,4 @@
+#from https://github.com/victoresque/pytorch-template
 import logging
 import torch.nn as nn
 import numpy as np

base/base_trainer.py

@@ -1,3 +1,4 @@
+#from https://github.com/victoresque/pytorch-template, with heavy modifications
 import os
 import math
 import json, copy

configs/cf_IAMslant_noMask_charSpecSingleAppend_GANMedMT_autoAEMoPrcp2tightNewCTCUseGen_balB_hCF0.75_sMG.json

@@ -73,7 +73,7 @@
  "metrics": [],
  "trainer": {
  "class": "HWWithStyleTrainer",
- "iterations": 200000,
+ "iterations": 175000,
  "save_dir": "saved/",
  "val_step": 10000,
  "save_step": 25000,

configs/cf_RIMESLines_hwr_cnnOnly_batchnorm_aug.json

@@ -44,7 +44,7 @@
  "metrics": [],
  "trainer": {
  "class": "HWWithStyleTrainer",
- "iterations": 500000,
+ "iterations": 425000,
  "save_dir": "../saved/",
  "val_step": 12500,
  "save_step": 25000,

configs/cf_RIMESLinesslant_noMask_charSpecSingleAppend_GANMedMT_autoAEMoPrcp2tightNewCTCUseGen_balB_hCF0.75_sMG.json

@@ -71,7 +71,7 @@
  "metrics": [],
  "trainer": {
  "class": "HWWithStyleTrainer",
- "iterations": 200000,
+ "iterations": 175000,
  "save_dir": "saved/",
  "val_step": -1,
  "save_step": 25000,