From 471df66955b2cc1e585b97d446171039b0a91bf4 Mon Sep 17 00:00:00 2001
From: kapoorlab <randomaccessiblekapoor@gmail.com>
Date: Tue, 22 Aug 2023 07:13:54 +0200
Subject: [PATCH] default pass

---
 src/oneat/NEATModels/neat_densevollnet.py |  7 ++-----
 src/oneat/NEATModels/neat_vollnet.py      |  8 ++------
 src/oneat/NEATUtils/utils.py              | 15 +++++++++++++++
 3 files changed, 19 insertions(+), 11 deletions(-)

diff --git a/src/oneat/NEATModels/neat_densevollnet.py b/src/oneat/NEATModels/neat_densevollnet.py
index d191cf0..9b8a37b 100644
--- a/src/oneat/NEATModels/neat_densevollnet.py
+++ b/src/oneat/NEATModels/neat_densevollnet.py
@@ -1,7 +1,7 @@
 from oneat.NEATUtils import plotters
 import numpy as np
 from oneat.NEATUtils import utils
-from oneat.NEATUtils.utils import save_volume, pad_timelapse, get_nearest_volume,  load_json, volumeyoloprediction, normalizeFloatZeroOne, GenerateVolumeMarkers, MakeForest,save_volume_csv, volume_dynamic_nms
+from oneat.NEATUtils.utils import save_volume, create_sub_image, pad_timelapse, get_nearest_volume,  load_json, volumeyoloprediction, normalizeFloatZeroOne, GenerateVolumeMarkers, MakeForest,save_volume_csv, volume_dynamic_nms
 from keras import callbacks
 import os
 import sys
@@ -314,10 +314,7 @@ def predict(self,
            print(f'zero padded image shape ${self.image.shape}')
            self.second_pass_predict()
         if self.remove_markers == None:
-           self.pad_width = (self.config['imagey'], self.config['imagex']) 
-           self.image = np.zeros([self.originalimage.shape[0], self.originalimage.shape[1],  self.originalimage.shape[2] + self.pad_width[0], self.originalimage.shape[3] + self.pad_width[1] ]) 
-           for i in range(self.originalimage.shape[0]):
-              self.image[i,:] = pad_timelapse(self.originalimage[i,:], self.pad_width)
+           self.image = create_sub_image(self.originalimage,self.config['imagez'],self.config['imagey'], self.config['imagex']) 
            self.default_pass_predict() 
 
    
diff --git a/src/oneat/NEATModels/neat_vollnet.py b/src/oneat/NEATModels/neat_vollnet.py
index e11f588..a056042 100644
--- a/src/oneat/NEATModels/neat_vollnet.py
+++ b/src/oneat/NEATModels/neat_vollnet.py
@@ -1,7 +1,7 @@
 from oneat.NEATUtils import plotters
 import numpy as np
 from oneat.NEATUtils import utils
-from oneat.NEATUtils.utils import save_volume, pad_timelapse, get_nearest_volume,  load_json, volumeyoloprediction, normalizeFloatZeroOne, GenerateVolumeMarkers, MakeForest,save_volume_csv, volume_dynamic_nms
+from oneat.NEATUtils.utils import save_volume, create_sub_image,  pad_timelapse, get_nearest_volume,  load_json, volumeyoloprediction, normalizeFloatZeroOne, GenerateVolumeMarkers, MakeForest,save_volume_csv, volume_dynamic_nms
 from keras import callbacks
 import os
 import sys
@@ -311,12 +311,8 @@ def predict(self,
            print(f'zero padded image shape ${self.image.shape}')
            self.second_pass_predict()
         if self.remove_markers == None:
-           self.pad_width = (self.config['imagey'], self.config['imagex']) 
-           self.image = np.zeros([self.originalimage.shape[0], self.originalimage.shape[1],  self.originalimage.shape[2] + self.pad_width[0], self.originalimage.shape[3] + self.pad_width[1] ]) 
            
-           for i in range(self.originalimage.shape[0]):
-              self.image[i,:] = pad_timelapse(self.originalimage[i,:], self.pad_width) 
-           self.image = self.originalimage
+           self.image = create_sub_image(self.originalimage,self.config['imagez'],self.config['imagey'], self.config['imagex']) 
            self.default_pass_predict() 
 
    
diff --git a/src/oneat/NEATUtils/utils.py b/src/oneat/NEATUtils/utils.py
index 864702a..d790e65 100644
--- a/src/oneat/NEATUtils/utils.py
+++ b/src/oneat/NEATUtils/utils.py
@@ -398,6 +398,21 @@ def load_full_training_data(directory, filename, axes=None, verbose=True):
 
     return (X, Y), axes
 
+def create_sub_image(image, n, m, p):
+    
+    t, z, y, x = image.shape
+    
+    z_remainder = z % n
+    y_remainder = y % m
+    x_remainder = x % p
+    
+    new_z = z - z_remainder
+    new_y = y - y_remainder
+    new_x = x - x_remainder
+    
+    sub_image = image[:, :new_z, :new_y, :new_x]
+    
+    return sub_image
 
 def pad_timelapse(image, pad_width):