Newer model, better images

Author: code2k13

model.py

@@ -1,6 +1,6 @@
 import tensorflow as tf 
 
-IMG_SIZE = 1024
+IMG_SIZE = 512
 OUTPUT_CHANNELS = 1
 
 def downsample(filters, size, apply_batchnorm=True,strides = 2,name=''):

removestars.py

@@ -14,9 +14,9 @@
 
 
 IMG_SIZE = 512
-MODEL_SIZE = 1024
+MODEL_SIZE = 512
 all_outputs = 1
-pad_width = 2
+pad_width = 8
 total_steps = 0
 progress_bar = None
 current_progress= 0
@@ -35,9 +35,9 @@ def process_tile(channel,i,j,pad_width,model,output_image):
   current_tile = current_tile.resize((IMG_SIZE-pad_width*2,IMG_SIZE-pad_width*2))
   blank_image.paste(current_tile,(pad_width,pad_width))
   blank_image = blank_image.resize((MODEL_SIZE,MODEL_SIZE))
-  blank_image  = np.asarray(blank_image,dtype="float16").reshape(1,MODEL_SIZE,MODEL_SIZE)/512
+  blank_image  = np.asarray(blank_image,dtype="float32").reshape(1,MODEL_SIZE,MODEL_SIZE)/512
   predicted_section = model.predict(blank_image,verbose=0)      
-  predicted_section = predicted_section.reshape(MODEL_SIZE,MODEL_SIZE)*1024
+  predicted_section = predicted_section.reshape(MODEL_SIZE,MODEL_SIZE)*512
   predicted_section = Image.fromarray(predicted_section).convert('L')
   predicted_section =  predicted_section.resize((IMG_SIZE,IMG_SIZE))            
   predicted_section = predicted_section.crop((pad_width,pad_width,IMG_SIZE-pad_width,IMG_SIZE-pad_width))

requirements.txt

@@ -1,4 +1,5 @@
 tensorflow
 numpy
 pillow
-tqdm
+tqdm
+tensorflowjs

weights/weights.data-00000-of-00001

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b4e7b7a2066cc91cdf36d898dae5b98c0d005808c9e1906f640183ed116fb40
+oid sha256:985845db41f3949875762f4fa034b2a7a910ec4efde40086b8d6e52beef872fe
 size 124532216