Gigapixel image support & some error checking (#17)

**Bug fixes:**

* The `resize_data.py`, `find_bad.py`, & `vis.py` scripts now support gigapixel images. Hopefully the Pillow/PIL `DecompressionBombWarning` warning no longer shows up. python-pillow/Pillow#515, https://pillow.readthedocs.io/en/stable/reference/Image.html#PIL.Image.open You can manually change `Image.MAX_IMAGE_PIXELS = 1000000000` to `Image.MAX_IMAGE_PIXELS = None` if you're still running into issues.

* The `vis.py` script now checks that you specified the required number of image dimensions.


**Improvements:**

* General README improvements for color correlation matrices & color decorrelation.

README.md

@@ -125,6 +125,13 @@ python data_tools/calc_ms.py -data_path <training_data>
 
 Now you can start training your DeepDream model by running the GoogleNet training script. It's recommended that you save the model every 5-10 epochs in order to monitor the quality of the visualizations.
 
+After training your models, you can add a color correlation matrix to them for color decorrelation with the following command:
+
+
+```
+python data_tools/calc_cm.py -data_path <training_data> -model_file <bvlc_out120>.pth
+```
+
 
 ---
 

data_tools/README.md

@@ -9,7 +9,7 @@ All of these scripts with the exception of `sort_images.py` can be copied to and
 
    1. [Dataset Mean and Standard Deviation Calculation](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#dataset-mean-and-standard-deviation-calculation)
 
-   2. [Dataset RGB Covariance Matrix Calculation](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#dataset-rgb-covariance-matrix-calculation)
+   2. [Dataset Color Correlation Matrix Calculation](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#dataset-color-correlation-matrix-calculation)
 
    3. [FC Channel Contents](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#fc-channel-contents)
 
@@ -21,13 +21,13 @@ All of these scripts with the exception of `sort_images.py` can be copied to and
 
    7. [Automatic Image Sorter](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#automatic-image-sorter)
 
-2. [Model Tools]()
+2. [Model Tools](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#model-editing)
 
    1. [Reduce Model Size](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#reduce-model-size)
 
    2. [Add/Change Model Values](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#addchange-model-values)
 
-3. [Visualization & Training Tools]()
+3. [Visualization & Training Tools](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#comparison-of-results)
 
    1. [Graph Training Data](https://github.com/ProGamerGov/dream-creator/tree/master/data_tools#graph-training-data)
 
@@ -50,9 +50,9 @@ python calc_ms.py -data_path <training_data>
 * `-use_rgb`: Enabling this flag will result in output values being in RGB format instead of BGR.
 
 
-## Dataset RGB Covariance Matrix Calculation
+## Dataset Color Correlation Matrix Calculation
 
-This script calculates the RGB covariance matrix required for color decorrelation.
+This script calculates the color correlation matrix required for color decorrelation.
 
 ```
 python calc_cm.py -data_path <training_data>

data_tools/find_bad.py

@@ -3,6 +3,8 @@
 from PIL import Image
 
 
+Image.MAX_IMAGE_PIXELS = 1000000000 # Support gigapixel images
+
 
 def main():
     parser = argparse.ArgumentParser()

data_tools/resize_data.py

@@ -3,6 +3,9 @@
 from PIL import Image
 
 
+Image.MAX_IMAGE_PIXELS = 1000000000 # Support gigapixel images
+
+
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument("-data_path", help="Path to your dataset", type=str, default='')

utils/vis_utils.py

@@ -7,6 +7,9 @@
 from utils.inceptionv1_caffe import InceptionV1_Caffe
 
 
+Image.MAX_IMAGE_PIXELS = 1000000000 # Support gigapixel images
+
+
 def calc_image_size(image_path, image_size):
     image = Image.open(image_path).convert('RGB')
     if type(image_size) is not tuple and type(image_size) is not list or len(image_size) == 1:

vis.py

@@ -60,6 +60,9 @@ def main():
 def main_func(params):
     if params.content_image != '':
         params.image_size = calc_image_size(params.content_image, params.image_size)
+    else:
+        assert len(params.image_size) > 1, "two -image_size values are required when not using a content image"
+
     if params.seed > -1:
         set_seed(params.seed)