minor corrections

Author: solivr

INSTALL.md

@@ -3,7 +3,7 @@
 
 - Create a virtual environment with all the packages : `conda env create -f environment.yml`
 
-- Then activate the environment with `source activate doc_seg`
+- Then activate the environment with `source activate dh_segment`
 
 - It might be possible that the following needs to be added to your `~/.bashrc`
 

README.md

@@ -24,10 +24,17 @@ In order to limit memory usage, the images in the dataset we provide have been d
 __How to__
 
 1. Get the annotated dataset [here](https://github.com/dhlab-epfl/dhSegment/releases/download/v0.2/pages.zip), which already contains the folders `images` and `labels` for training validation and testing set. Unzip it into `model/pages`. 
-2. You can train the model from scratch with 
+2. Download the pretrained weights for ResNet :
+```
+cd pretrained_models/
+python download_resnet_pretrained_model.py
+cd ..
+```
+
+3. You can train the model from scratch with 
     `python train.py with demo/demo_config.json`
     or skip this step and use directly the [provided model](https://github.com/dhlab-epfl/dhSegment/releases/download/v0.2/model.zip) (download and unzip it in `demo/model`)
-3. Run `python demo.py`
-4. Have a look at the results in `demo/processed_images`
+4. Run `python demo.py`
+5. Have a look at the results in `demo/processed_images`
 
 

demo.py

@@ -8,7 +8,6 @@
 import numpy as np
 import os
 import cv2
-import argparse
 import tempfile
 from scipy.misc import imread, imsave
 
@@ -92,7 +91,7 @@ def find_page(img_filenames, dir_predictions, output_dir):
             target_shape = (orig_img.shape[1], orig_img.shape[0])
             bin_upscaled = cv2.resize(np.uint8(page_bin), target_shape, interpolation=cv2.INTER_NEAREST)
 
-            # Find quadrilateral inclosing the page
+            # Find quadrilateral enclosing the page
             pred_box = boxes_detection.find_box(np.uint8(bin_upscaled), mode='quadrilateral')
 
             if pred_box is not None:

environment.yml

@@ -1,4 +1,4 @@
-name: doc_seg
+name: dh_segment
 channels:
 - defaults
 dependencies:

pretrained_models/download_vgg_pretrained_model.py

@@ -0,0 +1,39 @@
+#!/usr/bin/env python
+
+import urllib.request
+import tarfile
+import os
+from tqdm import tqdm
+
+
+def progress_hook(t):
+    last_b = [0]
+
+    def update_to(b=1, bsize=1, tsize=None):
+        """
+        b  : int, optional
+            Number of blocks transferred so far [default: 1].
+        bsize  : int, optional
+            Size of each block (in tqdm units) [default: 1].
+        tsize  : int, optional
+            Total size (in tqdm units). If [default: None] remains unchanged.
+        """
+        if tsize is not None:
+            t.total = tsize
+        t.update((b - last_b[0]) * bsize)
+        last_b[0] = b
+
+    return update_to
+
+
+if __name__ == '__main__':
+    tar_filename = 'vgg_16.tar.gz'
+    with tqdm(unit='B', unit_scale=True, unit_divisor=1024, miniters=1,
+              desc="Downloading pre-trained weights") as t:
+        urllib.request.urlretrieve('http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz', tar_filename,
+                                   reporthook=progress_hook(t))
+    tar = tarfile.open(tar_filename)
+    tar.extractall()
+    tar.close()
+    print('VGG-16 pre-trained weights downloaded!')
+    os.remove(tar_filename)