fix some spelling and grammar errors

dlib/algs.h

@@ -74,7 +74,7 @@
 
 // DNN module uses template-based network declaration that leads to very long
 // type names. Visual Studio will produce Warning C4503 in such cases. https://msdn.microsoft.com/en-us/library/074af4b6.aspx says
-// that correct binaries are still produced even when this warning happens, but linker errors from visual studio, if they occurr could be confusing.
+// that correct binaries are still produced even when this warning happens, but linker errors from visual studio, if they occur could be confusing.
 #pragma warning( disable: 4503 )
 
 

dlib/bigint/bigint_kernel_abstract.h

@@ -42,7 +42,7 @@ namespace dlib
                 operator--(int)  // post decrement
 
 
-                the other comparason operators(>, !=, <=, and >=) are 
+                the other comparison operators(>, !=, <=, and >=) are 
                 available and come from the templates in dlib::relational_operators
 
             THREAD SAFETY

dlib/disjoint_subsets/disjoint_subsets_sized_abstract.h

@@ -24,7 +24,7 @@ namespace dlib
                 of finding which subset a particular integer belongs to as well as
                 merging subsets.  It also allows you to find out how big each subset is.  It 
                 is therefore essentially the same thing as dlib::disjoint_subsets, except
-                it also keeps track of of the size of each subset.
+                it also keeps track of the size of each subset.
         !*/
     public:
 

dlib/gui_core/gui_core_kernel_abstract.h

@@ -121,7 +121,7 @@ namespace dlib
                     present in the pixel at all and 255 indicating that the color
                     is present in the pixel with maximum intensity.
 
-                    Note that the structure, order, and size of of this struct are 
+                    Note that the structure, order, and size of this struct are 
                     implementation dependent.  It will always contain fields called 
                     red, green, and blue but they may not be in that order and there 
                     may be padding.  

dlib/gui_widgets/base_widgets.cpp

@@ -562,7 +562,7 @@ namespace dlib
                 // this case here covers the unlikly event that you click on a button,
                 // move the mouse off the button and then move it back very quickly and
                 // release the mouse button.   It is possible that this mouse up event
-                // will occurr before any mouse move event so you might not have set
+                // will occur before any mouse move event so you might not have set
                 // that the button is depressed yet.
                 
                 // So we should say that this triggers an on_button_down() event and

dlib/gui_widgets/widgets.h

@@ -2842,7 +2842,7 @@ namespace dlib
                 {
                     rgb_alpha_pixel color;
                     assign_pixel(color, graph_.node(i).data.color);
-                    // this node is in area so lets draw it and all of it's edges as well
+                    // this node is in area so lets draw it and all of its edges as well
                     draw_solid_circle(c,center,rad-3,color,area);
                     color.alpha = 240;
                     draw_circle(c,center,rad-3,color,area);

dlib/image_keypoint/binned_vector_feature_image_abstract.h

@@ -39,7 +39,7 @@ namespace dlib
                 That is, the output vector has a dimensionality that is equal to the number
                 of hash bins times the dimensionality of the lower level vector plus one.
                 The value in the extra dimension concatenated onto the end of the vector is
-                always a constant value of of 1 and serves as a bias value.  This means
+                always a constant value of 1 and serves as a bias value.  This means
                 that, if there are N hash bins, these vectors are capable of representing N
                 different linear functions, each operating on the vectors that fall into
                 their corresponding hash bin.

dlib/image_transforms/interpolation_abstract.h

@@ -428,7 +428,7 @@ namespace dlib
               dlib/image_processing/generic_image.h 
             - pixel_traits<typename image_traits<image_type>::pixel_type>::has_alpha == false
         ensures
-            - Resizes img so that each of it's dimensions are size_scale times larger than img.
+            - Resizes img so that each of its dimensions are size_scale times larger than img.
               In particular, we will have:
                 - #img.nr() == std::round(size_scale*img.nr())
                 - #img.nc() == std::round(size_scale*img.nc())
@@ -1035,7 +1035,7 @@ namespace dlib
                 - #angle == 0
                 - #rows and #cols is set such that the total size of the chip is as close
                   to size_ as possible but still matches the aspect ratio of rect_.
-                - As long as size_ and the aspect ratio of of rect_ stays constant then
+                - As long as size_ and the aspect ratio of rect_ stays constant then
                   #rows and #cols will always have the same values.  This means that, for
                   example, if you want all your chips to have the same dimensions then
                   ensure that size_ is always the same and also that rect_ always has the
@@ -1057,7 +1057,7 @@ namespace dlib
                 - #angle == angle_
                 - #rows and #cols is set such that the total size of the chip is as close
                   to size_ as possible but still matches the aspect ratio of rect_.
-                - As long as size_ and the aspect ratio of of rect_ stays constant then
+                - As long as size_ and the aspect ratio of rect_ stays constant then
                   #rows and #cols will always have the same values.  This means that, for
                   example, if you want all your chips to have the same dimensions then
                   ensure that size_ is always the same and also that rect_ always has the

dlib/lsh/create_random_projection_hash_abstract.h

@@ -42,7 +42,7 @@ namespace dlib
                   numbers and we also perform basic centering to ensure the plane passes
                   though the data.
             - This function uses the supplied random number generator, rnd, to drive part
-              of it's processing.  Therefore, giving different random number generators
+              of its processing.  Therefore, giving different random number generators
               will produce different outputs.
     !*/
 
@@ -108,7 +108,7 @@ namespace dlib
                   We train it on randomly selected and randomly labeled points from v.
                   The C SVM parameter is set to the given C argument.
             - This function uses the supplied random number generator, rnd, to drive part
-              of it's processing.  Therefore, giving different random number generators
+              of its processing.  Therefore, giving different random number generators
               will produce different outputs.
     !*/
 

dlib/statistics/statistics_abstract.h

@@ -165,7 +165,7 @@ namespace dlib
                   were independent events.  The larger the magnitude of COR the more
                   evidence we have for the correlation.
                 - COR < 0: There is evidence that A and B are anti-correlated.  That is,
-                  when A happens B is unlikely to happen and vise versa.  The larger the
+                  when A happens B is unlikely to happen and vice versa.  The larger the
                   magnitude of COR the more evidence we have for the anti-correlation.
             - This function implements the simple likelihood ratio test discussed in the
               following paper:

dlib/svm/structural_svm_distributed_abstract.h

@@ -144,7 +144,7 @@ namespace dlib
 
                   Also note that sample risk is an upper bound on a sample's loss.  So
                   you can think of this epsilon value as saying "solve the optimization
-                  problem until the average loss per sample is within epsilon of it's 
+                  problem until the average loss per sample is within epsilon of its 
                   optimal value".
         !*/
 

dlib/svm/structural_svm_problem_abstract.h

@@ -123,7 +123,7 @@ namespace dlib
 
                   Also note that sample risk is an upper bound on a sample's loss.  So
                   you can think of this epsilon value as saying "solve the optimization
-                  problem until the average loss per sample is within epsilon of it's 
+                  problem until the average loss per sample is within epsilon of its 
                   optimal value".
         !*/
 

docs/docs/imaging.xml

@@ -692,7 +692,7 @@
                 That is, the output vector has a dimensionality that is equal to the number
                 of hash bins times the dimensionality of the lower level vector plus one.
                 The value in the extra dimension concatenated onto the end of the vector is
-                always a constant value of of 1 and serves as a bias value.  This means
+                always a constant value of 1 and serves as a bias value.  This means
                 that, if there are N hash bins, these vectors are capable of representing N
                 different linear functions, each operating on the vectors that fall into
                 their corresponding hash bin.

tools/htmlify/to_xml.cpp

@@ -718,7 +718,7 @@ void process_file (
                                 if (class_stack.size() > 0 && namespaces.back() == class_stack.top().name)
                                 {
                                     // If this class is a inner_class of another then push it into the
-                                    // public_inner_classes or protected_inner_classes field of it's containing class.
+                                    // public_inner_classes or protected_inner_classes field of its containing class.
                                     if (class_stack.size() > 1)
                                     {
                                         tok_class_record temp = class_stack.top();

tools/python/src/image2.cpp

@@ -247,7 +247,7 @@ void register_extract_image_chip (py::module& m)
     - self.angle == 0 \n\
     - self.rows and self.cols is set such that the total size of the chip is as close \n\
       to size as possible but still matches the aspect ratio of rect. \n\
-    - As long as size and the aspect ratio of of rect stays constant then \n\
+    - As long as size and the aspect ratio of rect stays constant then \n\
       self.rows and self.cols will always have the same values.  This means \n\
       that, for example, if you want all your chips to have the same dimensions \n\
       then ensure that size is always the same and also that rect always has \n\
@@ -261,7 +261,7 @@ void register_extract_image_chip (py::module& m)
                 - self.angle == 0
                 - self.rows and self.cols is set such that the total size of the chip is as close
                   to size as possible but still matches the aspect ratio of rect.
-                - As long as size and the aspect ratio of of rect stays constant then
+                - As long as size and the aspect ratio of rect stays constant then
                   self.rows and self.cols will always have the same values.  This means
                   that, for example, if you want all your chips to have the same dimensions
                   then ensure that size is always the same and also that rect always has
@@ -278,7 +278,7 @@ void register_extract_image_chip (py::module& m)
     - self.angle == angle \n\
     - self.rows and self.cols is set such that the total size of the chip is as \n\
       close to size as possible but still matches the aspect ratio of rect. \n\
-    - As long as size and the aspect ratio of of rect stays constant then \n\
+    - As long as size and the aspect ratio of rect stays constant then \n\
       self.rows and self.cols will always have the same values.  This means \n\
       that, for example, if you want all your chips to have the same dimensions \n\
       then ensure that size is always the same and also that rect always has \n\
@@ -292,7 +292,7 @@ void register_extract_image_chip (py::module& m)
                 - self.angle == angle
                 - self.rows and self.cols is set such that the total size of the chip is as
                   close to size as possible but still matches the aspect ratio of rect.
-                - As long as size and the aspect ratio of of rect stays constant then
+                - As long as size and the aspect ratio of rect stays constant then
                   self.rows and self.cols will always have the same values.  This means
                   that, for example, if you want all your chips to have the same dimensions
                   then ensure that size is always the same and also that rect always has