Version History: Difference between revisions
Jump to navigation
Jump to search
| Line 10: | Line 10: | ||
===Change Log=== | ===Change Log=== | ||
====Enhancements==== | ====Enhancements==== | ||
=====Compares to 3.0.0-rc1===== | |||
* Based on Open CV 3.0.0 release. For details, please see [http://opencv.org/opencv-3-0.html Open CV 3.0 release note] | * Based on Open CV 3.0.0 release. For details, please see [http://opencv.org/opencv-3-0.html Open CV 3.0 release note] | ||
* All versions of Emgu CV except the CUDA builds now use a single unmanaged dll (cvextern.dll) instead of the multiple dlls in prior releases (cvextern.dll, opencv_core{xxx}.dll, opencv_imgproc{xxx}.dll, etc). We are able to dramatically reduce the size of the dll compares to the old multi-file releases. For example, 3.0.0 open source final release, x86, non-cuda unmanaged binary contains a single cvextern.dll that is ~10MB in size. While the same build in 2.4.10 open source final release contains 19 files that is ~18MB in size. All CUDA build will still use the multi-file deployment model. We would like to make our CUDA build a single-dll release but we are having technical difficulty linking against static library that contains CUDA device code (The resulting binary will always cause Stack Overflow when CUDA code are called). | * All versions of Emgu CV except the CUDA builds now use a single unmanaged dll (cvextern.dll) instead of the multiple dlls in prior releases (cvextern.dll, opencv_core{xxx}.dll, opencv_imgproc{xxx}.dll, etc). We are able to dramatically reduce the size of the dll compares to the old multi-file releases. For example, 3.0.0 open source final release, x86, non-cuda unmanaged binary contains a single cvextern.dll that is ~10MB in size. While the same build in 2.4.10 open source final release contains 19 files that is ~18MB in size. All CUDA build will still use the multi-file deployment model. We would like to make our CUDA build a single-dll release but we are having technical difficulty linking against static library that contains CUDA device code (The resulting binary will always cause Stack Overflow when CUDA code are called). | ||
* Android version is now build with OpenMP support from Android NDK r10e. This enable multi-threading on mobile devices. Our Android demo app, such as Face Detection program now runs 4 times faster on a Quad-core android phone (tested on an HTC One M7). | * Android version is now build with OpenMP support from Android NDK r10e. This enable multi-threading on mobile devices. Our Android demo app, such as Face Detection program now runs 4 times faster on a Quad-core android phone (tested on an HTC One M7). | ||
* | |||
=====Highlights since 2.4.10===== | |||
* Windows Phone 8.1 support | |||
* Unity 5.0 support | |||
* Using Open CV's C++ interface | |||
* Open CV's T-API support | |||
* Optional parameters support in Emgu CV library | |||
* Emgu CV 3.0 can use OpenCL (GPU computing) on certain Android devices. A list of OpenCL supported Android phone can be found [http://arrayfire.com/opencl-on-mobile-devices/ here]. | |||
==Emgu.CV-3.0.0-rc1== | ==Emgu.CV-3.0.0-rc1== | ||
Revision as of 20:37, 10 June 2015
Emgu.CV-3.0.0
Coming soon...
Instruction to convert 2.4.x code to 3.0 can be found here
Online Documentation
Change Log
Enhancements
Compares to 3.0.0-rc1
- Based on Open CV 3.0.0 release. For details, please see Open CV 3.0 release note
- All versions of Emgu CV except the CUDA builds now use a single unmanaged dll (cvextern.dll) instead of the multiple dlls in prior releases (cvextern.dll, opencv_core{xxx}.dll, opencv_imgproc{xxx}.dll, etc). We are able to dramatically reduce the size of the dll compares to the old multi-file releases. For example, 3.0.0 open source final release, x86, non-cuda unmanaged binary contains a single cvextern.dll that is ~10MB in size. While the same build in 2.4.10 open source final release contains 19 files that is ~18MB in size. All CUDA build will still use the multi-file deployment model. We would like to make our CUDA build a single-dll release but we are having technical difficulty linking against static library that contains CUDA device code (The resulting binary will always cause Stack Overflow when CUDA code are called).
- Android version is now build with OpenMP support from Android NDK r10e. This enable multi-threading on mobile devices. Our Android demo app, such as Face Detection program now runs 4 times faster on a Quad-core android phone (tested on an HTC One M7).
Highlights since 2.4.10
- Windows Phone 8.1 support
- Unity 5.0 support
- Using Open CV's C++ interface
- Open CV's T-API support
- Optional parameters support in Emgu CV library
- Emgu CV 3.0 can use OpenCL (GPU computing) on certain Android devices. A list of OpenCL supported Android phone can be found here.
Emgu.CV-3.0.0-rc1
Instruction to convert 2.4.x code to 3.0 can be found here
Online Documentation
Change Log
Enhancements
- Based on Open CV 3.0.0-rc1 release. For details, please see Open CV 3.0 rc1 release note
- Added Windows Phone 8.1 support
- Added Unity 5.0 support
Emgu.CV-3.0.0-alpha
Online Documentation
Change Log
Enhancements
- Based on Open CV 3.0.0-alpha release. For details, please see Open CV 3.0 alpha release note
- Use Open CV C++ interfaces instead of the C interface
- Support for Open CV's T-API