ac and rag

Author: CR10-dv

geospatial_learn/handyplots.py

@@ -119,11 +119,39 @@ def plt_confmat(trueVals, predVals, cmap = plt.cm.gray, fmt="%d"):
     The confusion matrix and a plot
     """
     labels = np.unique(trueVals)
+    # the above heatmap function is used to create the plot
 
     skplt.metrics.plot_confusion_matrix(trueVals, predVals, normalize=True)
 
     conf = confusion_matrix(trueVals, predVals)
 
+    
+#    ax = plt.gca()
+#    # to be used for confusion matrix
+#
+#    ax.invert_yaxis()
+#    #ax.set_anchor('NW')
+#    #ax.invert_yaxis()
+#    # plot the mean cross-validation scores
+#    img = ax.pcolor(conf, cmap=cmap, vmin=None, vmax=None)
+#    img.update_scalarmappable()
+#    ax.set_xlabel('True')
+#    ax.set_ylabel('Predicted')
+#    ax.set_xticks(np.arange(len(labels)) + .5)
+#    ax.set_yticks(np.arange(len(labels)) + .5)
+#    ax.set_xticklabels(labels)
+#    ax.set_yticklabels(labels)
+#
+#    ax.set_aspect(1)
+#
+#    for p, color, value in zip(img.get_paths(), img.get_facecolors(), img.get_array()):
+#        x, y = p.vertices[:-2, :].mean(0)
+#        if np.mean(color[:3]) > 0.5:
+#            c = 'k'
+#        else:
+#            c = 'w'
+#        ax.text(x, y, fmt % value, color=c, ha="center", va="center")
+    
     return
     conf
 
@@ -218,7 +246,30 @@ def plot3d(data, features, feature_names, point_color = 0):
 
     plt.show()   
 
+    
+def plot_S2_cloud(user, paswd, start_date, end_date, aoiJson, print_date=True):
+    """
+    
+    Plot Sentinel 2 cloud percentage for a given time period 
+    
+    Parameters
+    ------------------
+    
+    data : np array
+        the aformentioned array of features
+    
+    features : list 
+        a list of feature indices, eg [1,2,3] or [4,3,1] 
+    
+    feature_names : list of strings
+        a list of feature names ef ['red', 'green', 'blue']
+        
+    """
+    dF, products = data.sent2_query(user, paswd, aoiJson, start_date, end_date)
+    
+    dF.sort_values('ingestiondate', inplace=True)
 
+    dF.plot.line('ingestiondate', 'cloudcoverpercentage')
 
 def plot_change(inArray):
     """ This assumes that features are column wise and rows are samples

geospatial_learn/raster.py

@@ -214,7 +214,7 @@ def tile_rasters(inImage, outputImage, tilesize):
     exit_codes = [p.wait() for p in procList]
             #print(i)
 
-def batch_translate(folder, wildcard, FMT=None):
+def batch_translate(folder, wildcard, FMT='Gtiff'):
     """ 
     Using the gdal python API, this function translates the format of files
     to commonly used formats
@@ -243,7 +243,7 @@ def batch_translate(folder, wildcard, FMT=None):
     if FMT == 'Gtiff':
         fmt = '.tif'    
 
-    fileList = glob2.glob(path.join(folder,'**', '**','*', wildcard))
+    fileList = glob2.glob(path.join(folder,'*'+wildcard))
     outList = list()
     files = np.arange(len(fileList))
 

geospatial_learn/setup.py

@@ -0,0 +1,8 @@
+from setuptools import setup
+from Cython.Build import cythonize
+
+setup(
+    name='Hello world app',
+    ext_modules=cythonize("hello.pyx"),
+    zip_safe=False,
+)

geospatial_learn/utilities.py

@@ -374,25 +374,36 @@ def ms_toposeg(inRas, outShp, iterations=100, algo='ACWE', band=2, dist=30,
 
     if usemin == True:
         minIm = peak_local_max(invert(img), min_distance=dist, indices=False)
-            
-
-
-    ste = selem.square(se)
-    dilated = binary_dilation(maxIm, selem=ste)
-    seg, _ = ndi.label(dilated)
-    cnt = list(np.unique(seg))
-    
-    cnt.pop(0)
-    #levelsets = [seg==s for s in cnt]
-    
-    iters = np.arange(iterations)
-    
-    orig = seg>0
+        
+    if algo=='ACWE':
+        ste = selem.square(se)
+        dilated = binary_dilation(maxIm, selem=ste)
+        seg, _ = ndi.label(dilated)
+        cnt = list(np.unique(seg))
+        
+        cnt.pop(0)
+        #levelsets = [seg==s for s in cnt]
+        
+        iters = np.arange(iterations)
+        
+        orig = seg>0
 #TODO - get fuse burner algo in this
 
     if algo=='GAC':
 
+        ste = selem.square(se)
         gimg = inverse_gaussian_gradient(img)
+        maxIm = peak_local_max(gimg, min_distance=dist, indices=False)
+        dilated = binary_dilation(maxIm, selem=ste)
+        seg, _ = ndi.label(dilated)
+        cnt = list(np.unique(seg))
+    
+        cnt.pop(0)
+        #levelsets = [seg==s for s in cnt]
+        
+        iters = np.arange(iterations)
+    
+        orig = seg>0
 
 
 
@@ -413,6 +424,47 @@ def ms_toposeg(inRas, outShp, iterations=100, algo='ACWE', band=2, dist=30,
             orig[bw==1]=1
             del inv, sk
 
+        if usemin==True:
+            
+            minIm = peak_local_max(invert(gimg), min_distance=dist, indices=False)           
+            dilated = binary_dilation(minIm, selem=ste)
+            seg, _ = ndi.label(dilated)
+            cnt = list(np.unique(seg))
+        
+            cnt.pop(0)        
+            iters = np.arange(iterations)    
+            orig = seg>0                
+            e2 = mh.bwperim(bw)
+            edge[e2==1]=1
+    
+            
+            if init != None:
+                initBw = orig
+                orig = mcv(img, iterations=init,init_level_set=initBw, 
+                           smoothing=smooth, lambda1=1,
+                    lambda2=1)
+                orig = orig>0
+            
+            for i in tqdm(iters):
+                    inv = invert(orig)
+                    sk = skeletonize(inv) 
+                    sk = _skelprune(sk)
+                    bw2 = mcv(img, iterations=1,init_level_set=orig, smoothing=smooth, lambda1=1,
+                        lambda2=1)
+                    bw2[sk==1]=0
+                    if useedge == True:
+                        bw2[edge==1]=0
+                    # why do this? I think seg=bw will result in a pointer....
+                    orig = np.zeros_like(bw2, dtype=np.bool)
+                    orig[bw2==1]=1
+        
+                    del inv, sk
+
+            bw[bw2==1]=1        
+            newseg, _ = nd.label(bw)  
+            
+
+            
     else:
         # let it run for a bit to avoid over seg
         if init != None:
@@ -438,48 +490,51 @@ def ms_toposeg(inRas, outShp, iterations=100, algo='ACWE', band=2, dist=30,
 
             del inv, sk
 
-    if usemin==True:
-        
-        
+     
+            
+            
+        if usemin==True:
+            
+            
+            
+            dilated = binary_dilation(minIm, selem=ste)
+            seg, _ = ndi.label(dilated)
+            cnt = list(np.unique(seg))
 
-        dilated = binary_dilation(minIm, selem=ste)
-        seg, _ = ndi.label(dilated)
-        cnt = list(np.unique(seg))
+            cnt.pop(0)        
+            iters = np.arange(iterations)    
+            orig = seg>0                
+            e2 = mh.bwperim(bw)
+            edge[e2==1]=1
 
-        cnt.pop(0)        
-        iters = np.arange(iterations)    
-        orig = seg>0                
-        e2 = mh.bwperim(bw)
-        edge[e2==1]=1
-
-        
-        if init != None:
-            initBw = orig
-            orig = mcv(img, iterations=init,init_level_set=initBw, 
-                       smoothing=smooth, lambda1=1,
-                lambda2=1)
-            orig = orig>0
-        
-        for i in tqdm(iters):
-                inv = invert(orig)
-                sk = skeletonize(inv) 
-                sk = _skelprune(sk)
-                bw2 = mcv(img, iterations=1,init_level_set=orig, smoothing=smooth, lambda1=1,
+            
+            if init != None:
+                initBw = orig
+                orig = mcv(img, iterations=init,init_level_set=initBw, 
+                           smoothing=smooth, lambda1=1,
                     lambda2=1)
-                bw2[sk==1]=0
-                if useedge == True:
-                    bw2[edge==1]=0
-                # why do this? I think seg=bw will result in a pointer....
-                orig = np.zeros_like(bw2, dtype=np.bool)
-                orig[bw2==1]=1
-    
-                del inv, sk
-#    remove_small_objects(bw, min_size=3, in_place=True)
-        bw[bw2==1]=1        
-        newseg, _ = nd.label(bw)        
+                orig = orig>0
 
-    else:
-        newseg, _ = nd.label(bw)
+            for i in tqdm(iters):
+                    inv = invert(orig)
+                    sk = skeletonize(inv) 
+                    sk = _skelprune(sk)
+                    bw2 = mcv(img, iterations=1,init_level_set=orig, smoothing=smooth, lambda1=1,
+                        lambda2=1)
+                    bw2[sk==1]=0
+                    if useedge == True:
+                        bw2[edge==1]=0
+                    # why do this? I think seg=bw will result in a pointer....
+                    orig = np.zeros_like(bw2, dtype=np.bool)
+                    orig[bw2==1]=1
+        
+                    del inv, sk
+#    remove_small_objects(bw, min_size=3, in_place=True)
+            bw[bw2==1]=1        
+            newseg, _ = nd.label(bw)        
+                
+        else:
+            newseg, _ = nd.label(bw)
 
     if close==True:
         ste2 = selem.square(3)