Commit fc486996 authored by Davis King's avatar Davis King

updated release notes

parent 94254c45
...@@ -12,43 +12,41 @@ ...@@ -12,43 +12,41 @@
<current> <current>
New Features: New Features:
- Added scan_image_custom - Added routines for performing BFGS and L-BFGS optimization with box constraints.
- Added split_array() See the new find_min_box_constrained() and find_max_box_constrained() routines.
- Added add_image_left_right_flips() - Added vector_normalizer_frobmetric. This is a tool for learning a
- Updated the interface to allow the user to set different loss values for Mahalanobis distance metric.
false alarming vs getting a correct detection. This is for the - The user can now set different loss values for false alarming vs getting a
structural_sequence_segmentation_trainer correct detection when using the structural_sequence_segmentation_trainer.
- Added an overload of clamp() that lets you use matrix valued lower/upper bounds. - Added an overload of clamp() that lets you use matrix valued lower/upper bounds.
- Added a backtracking_line_search(), find_min_box_constrained() and find_max_box_constrained() - New image processing tools:
- Added vector_normalizer_frobmetric - Added the scan_image_custom object, split_array(), and add_image_left_right_flips().
- Added a function for computing Felzenszwalb's 31 channel HOG image - Added extract_fhog_features(), this is a function for computing
representation. This is extract_fhog_features(). Felzenszwalb's 31 channel HOG image representation.
Non-Backwards Compatible Changes: Non-Backwards Compatible Changes:
- Refactored the image pyramid code. Now there is just one templated object, - Refactored the image pyramid code. Now there is just one templated object called
pyramid_down and you give it the downsampling amount as a template pyramid_down and you give it the downsampling amount as a template argument. To make
argument. So you will have to update your code slightly. Use the old code work with this change use the following substitutions:
following substitutions: change pyramid_down to pyramid_down&lt;2&gt;
change pyramid_down to pyramid_down&lt;2&gt; change pyramid_down_3_2 to pyramid_down&lt;3&gt;
change pyramid_down_3_2 to pyramid_down&lt;3&gt; change pyramid_down_4_3 to pyramid_down&lt;4&gt;
change pyramid_down_4_3 to pyramid_down&lt;4&gt; change pyramid_down_5_4 to pyramid_down&lt;5&gt;
change pyramid_down_5_4 to pyramid_down&lt;5&gt;
Bug fixes:
Bug fixes:
Other:
Other: - Made the structural SVM solver slightly faster.
- Moved python C++ utility headers into dlib/python and updated #include - Moved the python C++ utility headers from tools/python/src into dlib/python.
statements to reflect this move. - The PNG loader is now able to load grayscale images with an alpha channel.
- Made the structural SVM solver's cache slightly more efficient - Removed checks that prevented users from using references to functions with the
- Made the PNG loader able to load in grayscale images with an alpha channel. optimization code and forced the use of function pointers. This was to avoid
- Removed a bunch of checks that prevented users from using references to triggering a bug in gcc 4.0. Since that compiler is no longer officially supported
functions with the optimization code and forced the use of function by dlib I've removed these checks to increase usability.
pointers. This was to avoid triggering a bug in gcc 4.0. Since that - Made resize_image() use bilinear interpolation by default and also added a special
compiler is no longer officially supported by dlib I've removed these version of it that is optimized for this case.
checks to increase usability. - Dlib's cmake files will now automatically link to the Intel MKL on MS Windows
- Made resize_image() use bilinear interpolation by default and also added a platforms if the MKL is installed.
special version of it that is optimized for this case.
</current> </current>
<!-- ************************************************************************************** --> <!-- ************************************************************************************** -->
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment