libopencv_videostab405-4.5.5-150400.1.28 >
A bwUvp9| Z ηDr$G!z ˫ oٯ4*zz1A::g9F; gIq >Fk
ȯ(_ojwH)G&,? ou p7 /0B5vEUf@K -
M1 >`
i6iis^BȲ"x _ 3r+< 5RɍYgMBׄ-Gvd/TzxXDtjB &4E ߚ5qc=. 7%Pm8~k2J*Al;ɻv5xiu
Dx
$wy A(PL%#7IEƹx\
8 > p C L ? < d + T ( 0 4
8 @
D ( 8 8 9 8 : 8 > @ F G H I X Y , \ T ] \ ^ b c I d e f l u v w t x | y z 8 C libopencv_videostab405 4.5.5 150400.1.28 Video stabilization libraries for OpenCV Video stabilization libraries for OpenCV bwTibs-power9-11 pSUSE Linux Enterprise 15 SUSE LLC BSD-3-Clause https://www.suse.com/ System/Libraries https://opencv.org/ linux ppc64le p bwSbwS0363fdfe684ba0eca587568ce43ada8d7b8dbd1730f3103a0fd60c72fb4f26ff libopencv_videostab.so.4.5.5 root root root root opencv-4.5.5-150400.1.28.src.rpm libopencv_videostab.so.405()(64bit) libopencv_videostab405 libopencv_videostab405(ppc-64) @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @
/sbin/ldconfig /sbin/ldconfig libc.so.6()(64bit) libc.so.6(GLIBC_2.17)(64bit) libgcc_s.so.1()(64bit) libgcc_s.so.1(GCC_3.0)(64bit) libm.so.6()(64bit) libm.so.6(GLIBC_2.17)(64bit) libm.so.6(GLIBC_2.27)(64bit) libm.so.6(GLIBC_2.29)(64bit) libopencv_calib3d.so.405()(64bit) libopencv_core.so.405()(64bit) libopencv_features2d.so.405()(64bit) libopencv_imgproc.so.405()(64bit) libopencv_photo.so.405()(64bit) libopencv_video.so.405()(64bit) libopencv_videoio.so.405()(64bit) libstdc++.so.6()(64bit) libstdc++.so.6(CXXABI_1.3)(64bit) libstdc++.so.6(GLIBCXX_3.4)(64bit) libstdc++.so.6(GLIBCXX_3.4.11)(64bit) libstdc++.so.6(GLIBCXX_3.4.21)(64bit) libstdc++.so.6(GLIBCXX_3.4.9)(64bit) rpmlib(CompressedFileNames) rpmlib(FileDigests) rpmlib(PayloadFilesHavePrefix) rpmlib(PayloadIsXz) 3.0.4-1 4.6.0-1 4.0-1 5.2-1 libopencv405 4.5.5-150400.1.28 4.14.3 a aa)@a,@`B`% @_`@__]@_^ _>e^^n@^&^%@^O@]@]@]fl]e@]Z@]N]M`@]G@[@[
@[ @ZkZS]@Z[@Z @ZYYZ@YYdYjY5GY5GY*W@WV@WEV@V@V'@V@VяVVzV^@VBU(U@U@T,@alarrosa@suse.com stefan.bruens@rwth-aachen.de dmueller@suse.com jengelh@inai.de stefan.bruens@rwth-aachen.de aloisio@gmx.com stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de dmueller@suse.com mliska@suse.cz stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de dimstar@opensuse.org stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de stefan.bruens@rwth-aachen.de marius.kittler@suse.com jengelh@inai.de jbrielmaier@suse.de christophe@krop.fr christophe@krop.fr stefan.bruens@rwth-aachen.de christophe@krop.fr fabian@ritter-vogt.de fvogt@suse.com schwab@suse.de tchvatal@suse.com stefan.bruens@rwth-aachen.de mpluskal@suse.com ecsos@opensuse.org ecsos@opensuse.org mpluskal@suse.com mpluskal@suse.com kah0922@gmail.com stefan.bruens@rwth-aachen.de tchvatal@suse.com tchvatal@suse.com stefan.bruens@rwth-aachen.de olaf@aepfle.de dimstar@opensuse.org martin.liska@suse.com toddrme2178@gmail.com toddrme2178@gmail.com tittiatcoke@gmail.com joerg.lorenzen@ki.tng.de joerg.lorenzen@ki.tng.de olaf@aepfle.de alarrosa@suse.com ohering@suse.de olaf@aepfle.de mlin@suse.com coolo@suse.com coolo@suse.com tittiatcoke@gmail.com - Remove the memoryperjob constraint which doesn't work as one
would expect and breaks ppc64 builds.
- Use %limit_memory -m 1700 to set the number of concurrent
jobs to a sane value and fix OOM errors when building in
workers with many cores.
- Decrease the disk constraint to 9G which seems to be enough - update to 4.5.5, highlights below, for details check
https://github.com/opencv/opencv/wiki/ChangeLog#version455
* Audio support as part of VideoCapture API: GStreamer #21264
* Updated SOVERSION handling rules: #21178
* DNN module patches:
+ Added tests to cover ONNX conformance test suite: #21088
+ Improved layers / activations / supported more models
+ Upgraded builtin protobuf from 3.5.2 to 3.19.1
+ More optimizations for RISC-V platform
+ Intel® Inference Engine backend ( OpenVINO™ ):
added support for OpenVINO 2021.4.2 LTS release
* G-API module:
+ G-API framework:
- Fixed issue with accessing 1D data from cv::RMat: #21103
- Restricted passing the G-API types to graph inputs/outputs
for execution: #21041
- Various fixes in G-API Doxygen reference: #20924
- Renamed various internal structures for consistency #20836 #21040
+ Fluid backend:
- Introduced a better vectorized version of Resize: #20664.
- Added vectorized version of Multiply kernel: #21024
- Added vectorized version of Divide kernel: #20914
- Added vectorized version of AddC kernel: #21119
- Added vectorized version of SubC kernel: #21158
- Added vectorized version of MulC kernel: #21177
- Added vectorized version of SubRC kernel: #21231
- Enabled SIMD dispatching for AbsDiffC: #21204
+ OpenCL backend:
- Fixed sporadic test failures in Multiply kernel running
on GPU: #21205
+ Intel® OpenVINO™ inference backend:
- Extended ie::Params to support static batch size as input
to inference: #20856
- Enabled 2D input tensor support in IE backend: #20925
- Fixed various issues with imported (pre-compiled)
networks: #20918
+ Media integration:
- Introduced a GStreamer-based pipeline source for
G-API: #20709
- Completed the integration of Intel® oneVPL as a pipeline
source for G-API #20773 with device selection #20738,
asynchronous execution #20901, intial demux support #21022,
and GPU-side memory allocation via DirectX 11 #21049.
+ Samples:
- Replaced custom kernels with now-standard G-API operations
in several samples #21106
- Moved API snippets from G-API samples to a dedicated
place #20857
+ Other changes and fixes:
- Fixed various static analysis issues for OpenVINO 2021.4
release: #21083 and #21212
- Fixed various build warnings introduced after OpenVINO
update: #20937
- Continued clean-up in the G-API test suite on GTest macros
[#20922] and test data #20995
- Added custom accuracy comparison functions to Fluid
performance tests: #21150.
* And many other contributions:
+ Added QRcode encoder: #17889
+ GSoC - OpenCV.js: Accelerate OpenCV.js DNN via WebNN: #20406
+ Add conventional Bayer naming: #20970
+ (opencv_contrib) Add Radon transform function to ximgproc: #3090
+ (opencv_contrib) New superpixel algorithm (F-DBSCAN): #3093
+ Created Stitching Tool: #21020
+ Improve CCL with new algorithms and tests: #21275
+ (opencv_contrib) Update ArUco tutorial: #3126
- Adjust memory constraints (mostly required for aarch64 on Leap)
- Add 0001-highgui-Fix-unresolved-OpenGL-functions-for-Qt-backe.patch - update to 4.5.4:
* 8-bit quantization in the dnn module
* Improved Julia bindings
* Speech recognition sample
* dnn module optimizations for RISC-V
* Tutorial about universal intrinsics and parallel_for usage
* Improvements in the dnn module:
- New layers and models support
- Some existing layers have been fixed
- Soft-NMS implementation
- Supported OpenVINO 2021.4.1 LTS release - Remove dependency on IlmBase, opencv never uses this directly. - update to 4.5.2, highlights below, for details check
https://github.com/opencv/opencv/wiki/ChangeLog#version452
* core: added support for parallel backends.
* imgproc: added IntelligentScissors implementation (JS demo).
* videoio: improved hardware-accelerated video de-/encoding tasks.
* DNN module:
+ Improved debugging of TensorFlow parsing errors: #19220
+ Improved layers / activations / supported more models:
- optimized: NMS processing, DetectionOutput
- fixed: Div with constant, MatMul, Reshape
(TensorFlow behaviour)
- added support: Mish ONNX subgraph, NormalizeL2 (ONNX),
LeakyReLU (TensorFlow), TanH + SAM (Darknet), Exp
+ Intel® Inference Engine backend ( OpenVINO™ ):
added support for OpenVINO 2021.3 release
* G-API module:
+ Python support:
- Introduced a new Python backend - now G-API can run custom
kernels written in Python as part of the pipeline: #19351
- Extended Inference support in the G-API bindings: #19318
- Added more graph data types in the G-API bindings: #19319
+ Inference support:
- Introduced dynamic input / CNN reshape functionality in the
OpenVINO inference backend #18240
- Introduced asynchronous execution support in the OpenVINO
inference backend, now it can run in multiple parallel
requests to increase stream density/throughput: #19487, #19425
- Extended supported data types with INT64/INT32 in ONNX
inference backend and with INT32 in the OpenVINO inference
backend #19792
- Introduced cv::GFrame / cv::MediaFrame and constant support
in the ONNX backend: #19070
+ Media support:
- Introduced cv::GFrame / cv::MediaFrame support in the
drawing/rendering interface: #19516
- Introduced multi-stream input support in Streaming mode
and frame synchronization policies to support cases like
Stereo: #19731
- Added Y and UV operations to access NV12 data of cv::GFrame
at the graph level; conversions are done on-the-fly if the
media format is different: #19325
+ Operations and kernels:
- Added performance tests for new operations (MorphologyEx,
BoundingRect, FitLine, FindContours, KMeans, Kalman,
BackgroundSubtractor)
- Fixed RMat input support in the PlaidML backend: #19782
- Added ARM NEON optimizations for Fluid AbsDiffC, AddWeighted,
and bitwise operations: #18466, #19233
- Other various static analysis and warning fixes
+ Documentation:
- [GSoC] Added TF/PyTorch classification conversion: #17604
- [GSoC] Added TF/PyTorch segmentation conversion: #17801
- [GSoC] Added TF/PyTorch detection model conversion: #18237
- Updated documentation to address Wide Universal Intrinsics
(WUI) SIMD API: #18952
+ And many other great contributions from OpenCV community:
- core: cuda::Stream constructor with stream flags: #19286
- highgui: pollKey() implementation for w32 backend: #19411
- imgcodecs: Added Exif parsing for PNG: #19439
- imgcodecs: OpenEXR compression options: #19540
- imgproc: connectedComponents optimizations: (Spaghetti
Labeling): #19631
- videoio: Android NDK camera support #19597
- (contrib) WeChat QRCode module open source: #2821
- (contrib) Implemented cv::cuda::inRange(): #2803
- (contrib) Added algorithms from Edge Drawing Library: #2313
- (contrib) Added Python bindings for Viz module: #2882
- Add libva build dependency for HW accelerated videoio
- Slight bump for memory constraints - Enable aruco module (recognize markers to detect camera pose) - update to 4.5.1, highlights below, for details check
https://github.com/opencv/opencv/wiki/ChangeLog#version451
* Continued merging of GSoC 2020 results:
+ Develop OpenCV.js DNN modules for promising web use cases
together with their tutorials
+ OpenCV.js: WASM SIMD optimization 2.0
+ High Level API and Samples for Scene Text Detection and
Recognition
+ SIFT: SIMD optimization of GaussianBlur 16U
* DNN module:
+ Improved layers / activations / supported more models:
- optimized: 1D convolution, 1D pool
- fixed: Resize, ReduceMean, Gather with multiple outputs,
importing of Faster RCNN ONNX model
- added support: INT32 ONNX tensors
+ Intel® Inference Engine backend (OpenVINO):
- added support for OpenVINO 2021.2 release
- added preview support for HDDL
+ Fixes and optimizations in DNN CUDA backend (thanks to
@YashasSamaga)
* G-API Framework:
+ Introduced serialization for cv::RMat, including
serialization for user-defined memory adapters
+ Introduced desync, a new Operation for in-graph asynchronous
execution - to allow different parts of the graph run with
a different latency
+ Introduced a notion of "in-graph metadata", now various
media-related information can be accessed in graph directly
(currently only limited to timestamps and frame IDs)
+ Introduced a new generic task-based executor, based on
Threading Building Blocks (TBB)
+ Extended infer<>() API to accept a new cv::GFrame data
structure to allow handling of various media formats without
changes in the graph structure
+ Made copy() an intrinsic where real copy may not happen
(optimized out) based on graph structure, extended it to
support cv::GFrame
+ Various fixes, including addressig static analysis,
documentation, and test issues
* G-API Operations:
+ Introduced new operations morphologyEx, boundingRect,
fitLine, kmeans, Background Subtractor, Kalman filter
* G-API Intel® Inference Engine backend (OpenVINO):
+ Extended cv::gapi::ie::Params<> to import CNN networks (e.g.
pre-compiled ones) instead of passing .XML and .BIN files;
also enabled configuring Inference Engine plugins via
this structure
+ Added a new overload to infer<>() to run inference over a
single region of interest
+ Added support for cv::MediaFrame input data type (projected
from cv::GFrame) and handling for NV12 input image format
* G-API Python bindings:
+ Exposed G-API's Inference and Streaming APIs in the OpenCV
Python bindings
+ Added initial Python support for cv::GArray data structure
* Significant progress on RISC-V port.
- Updated constraints, bump memory to 5 GB
- Cleaned up spec file - Split library package, move all libraries with external
dependencies (Qt5, ffmpeg, gstreamer) into separate packages
- Move haar and LBP cascades into separate package, pull in from
objdetect and face (detect) libraries. - update to 4.5.0, see
https://github.com/opencv/opencv/wiki/ChangeLog#version450
for details, highlights:
* OpenCV license has been changed to Apache 2 (OpenCV 3.x will
keep using BSD)
* GSoC is over, all projects were success and most of them have
already been merged. Optimizations for RISC-V, bindings for
Julia language, real-time single object tracking, improved SIFT
and others
* OpenJPEG is now used by default for JPEG2000
* Supported multiple OpenCL contexts
* Improvements in dnn module:
+ Support latest OpenVINO 2021.1 release
+ Tengine lite support for inference on ARM
+ Many fixes and optimizations in CUDA backend
* Added Python bindings for G-API module
* Multiple fixes and improvements in flann module
* Added Robot-World/Hand-Eye calibration function - update to 4.4.0:
* SIFT (Scale-Invariant Feature Transform) algorithm has been
moved to the main repository (patent on SIFT is expired)
* DNN module:
* State-of-art Yolo v4 Detector: #17148.
* onnx: Add support for Resnet_backbone
* EfficientDet models
* add text recognition sample / demo
* FlowNet2 optical flow
* Intel Inference Engine backend
* added support for OpenVINO 2020.3 LTS / 2020.4 releases
* support of NN Builder API is planned for removal in the next release
* Many fixes and optimizations in CUDA backend
* Obj-C / Swift bindings: #17165
* Julia bindings as part of ongoing GSoC project
* BIMEF: A Bio-Inspired Multi-Exposure Fusion Framework for Low-light Image Enhancement
* Enable Otsu thresholding for CV_16UC1 images
* Add Stroke Width Transform algorithm for Text Detection
* Planned migration on Apache 2 license for next releases
- remove opencv-includedir.patch (obsolete) - Use memoryperjob constraint instead of %limit_build macro. - Update to 4.3.0
* DNN module:
+ Improved layers / activations / supported more models:
- ONNX: LSTM, Broadcasting, Algebra over constants, Slice with
multiple inputs
- DarkNet: grouped convolutions, sigmoid, swish, scale_channels
- MobileNet-SSD v3: #16760
+ New samples / demos:
- Clothes parts segmentation and CP-VTON
- DaSiamRPN tracker
Intel® Inference Engine backend (OpenVINO™):
- added support for custom layers through nGraph OpenVINO
API: #16628
- nGraph OpenVINO API is used by default: #16746
+ Many fixes and optimizations in CUDA backend (thanks to
@YashasSamaga)
+ OPEN AI LAB team submitted the patch that accelerates OpenCV
DNN on ARM using their Tengine library
* G-API module:
+ Introduced a new graph-level data type GOpaque. This type
can be used to pass arbitrary user data types between G-API
nodes in the graph (supported for CPU/OpenCV backend only).
+ Introduced a way to declare G-API CPU (OpenCV) kernels in-place
+ Added a new sample "Privacy masking camera", combining Deep
Learning with traditional Image Processing (link)
+ Added more operations in the default library: WarpAffine,
WarpPerspective, NV12toGray.
* Performance improvements:
+ IPP-ICV library with CPU optimizations has been updated to
version 2020.0.0 Gold
+ SIMD intrinsics: integral, resize, (opencv_contrib) RLOF
implementation #2476
* And many other great contributions from OpenCV community:
+ (opencv_contrib) Computer Vision based Alpha Matting
(GSoC 2019) #2306
+ calib3d: findChessboardCornersSB improvements: #16625
+ calib3d: updated documentation for RT matrices: #16860
+ core: improved getNumberOfCPUs(): #16268
+ imgproc: new algorithm HOUGH_GRADIENT_ALT is added to
HoughCircles() function #16561. It has much better recall
and precision
+ imgcodecs: added initial support for OpenJPEG library
(version 2+): #16494
+ highgui(Qt): added Copy to clipboard: #16677
+ dnn: TensorFlow, Darknet and ONNX importers improvements
by @ashishkrshrivastava
+ (opencv_contrib) added rapid module for silhouette based 3D
object tracking: #2356
+ (opencv_contrib) SIFT detector is enabled by default due
patents expiration (without requirement of NONFREE build
option)
+ help materials: OpenCV Cheat Sheet in Python: #4875
* Changes that can potentially break compatibility:
+ image filtering functions throws exception on empty input
(voting results)
- Packaging changes:
* Stop mangling CMake diagnostic output, no dependency versions
end up in the packages anyway, drop opencv-build-compare.patch
* Set empty OPENCV_DOWNLOAD_TRIES_LIST, skip downloads even when
network is available during builds (e.g. local build).
* Drop upstream GLES patches:
+ 0001-Do-not-include-glx.h-when-using-GLES.patch
+ opencv-gles.patch - Disable Python 2 bindings for Tumbleweed. - Drop Jasper (i.e jpeg2k) support (boo#1130404, boo#1144260)
JasPer is unmaintained, CVEs are not being addressed (some issues
received patches submitted to the upstream github project, but are
not being merged, other CVEs are considered unfixable). openSUSE
follows other distros in dropping JasPer now (much later than
most others, incl. Debian). - Add webp build dependency to use system libwebp instead of bundled
one.
- Enable dispatch of AVX512 optimized code. - Update to 4.2.0
* DNN module:
+ Integrated GSoC project with CUDA backend: #14827
+ Intel® Inference Engine backend ( OpenVINO™ ):
- support for nGraph OpenVINO API (preview / experimental): #15537
* G-API module:
+ Enabled in-graph inference: #15090. Now G-API can express more
complex hybrid CV/DL algorithms;
- Intel® Inference Engine backend is the only available now,
support for DNN module will be added in the future releases.
+ Extended execution model with streaming support: #15216. Decoding,
image processing, inference, and post-processing are now pipelined
efficiently when processing a video stream with G-API.
+ Added tutorials covering these new features: Face analytics
pipeline and a sample Face beautification algorithm.
* Performance improvements:
+ SIMD intrinsics: StereoBM/StereoSGBM algorithms, resize, integral,
flip, accumulate with mask, HOG, demosaic, moments
+ Muti-threading: pyrDown
* And many other great patches from OpenCV community:
+ VideoCapture: video stream extraction (demuxing) through
FFmpeg backend.
+ VideoCapture: waitAny() API for camera input multiplexing
(Video4Linux through poll() calls).
+ (opencv_contrib) new algorithm Rapid Frequency Selective
Reconstruction (FSR): #2296 + tutorial.
+ (opencv_contrib) RIC method for sparse match interpolation: #2367.
+ (opencv_contrib) LOGOS features matching strategy: #2383.
* Breaking changes:
+ Disabled constructors for legacy C API structures.
+ Implementation of Thread Local Storage (TLS) has been improved to
release data from terminated threads. API has been changed.
+ Don't define unsafe CV_XADD implementation by default.
+ Python conversion rules of passed arguments will be updated in
next releases: #15915. - Limit build parallelism with limit_build, some ARM and PPC workers
have a high SMP/memory ratio and run out of memory otherwise.
- Apply memory constraints (3GB) to all architectures, avoid being
scheduled on very weak workers. - Update to 4.1.2
* DNN module:
+ Intel Inference Engine backend (OpenVINO):
- 2019R3 has been supported
- Support modern IE Core API
- New approach for custom layers management. Now all the OpenCV
layers fallbacks are implemented as IE custom layers which
helps to improve efficiency due less graph partitioning.
- High-level API which introduces dnn::Model class and set of
task-specific classes such dnn::ClassificationModel,
dnn::DetectionModel, dnn::SegmentationModel. It supports
automatic pre- and post-processing for deep learning networks.
* Performance improvements and platforms support:
+ MSA SIMD implementation has been contributed for MIPS platforms:
https://github.com/opencv/opencv/pull/15422
+ OpenCV.js optimization (threading and SIMD as part of GSoC
project): https://github.com/opencv/opencv/pull/15371
+ More optimizations using SIMD intrinsics: dotProd, FAST corners,
HOG, LK pyramid (VSX), norm, warpPerspective, etc
+ Fixed detection of Cascade Lake CPUs
* And many other great patches from OpenCV community:
+ GUI: support topmost window mode (Win32/COCOA):
https://github.com/opencv/opencv/pull/14872
+ Java: fix Mat.toString() for higher dimensions:
https://github.com/opencv/opencv/pull/15181
+ Implementation of colormap "Turbo"
https://github.com/opencv/opencv/pull/15388
+ QR-Code detection accuracy improvement:
https://github.com/opencv/opencv/pull/15356
+ GSoC: Add learning-based super-resolution module:
https://github.com/opencv/opencv_contrib/pull/2229 and
https://github.com/opencv/opencv_contrib/pull/2231
+ Detection accuracy improvement of the white marker aruco
corners: https://github.com/opencv/opencv_contrib/pull/2236
+ Added pattern generator tool for aruco:
https://github.com/opencv/opencv_contrib/pull/2250
+ and special thanks to @sturkmen72 for improvind and cleaning
up code of samples/tutorials
* Breaking changes:
+ fixed values thresholding accuracy in calcHist()
* Security fixes: CVE-2019-15939 (boo#1149742).
- Enable Graph API (G-API)
- Minor spec file cleanup - Include pkg-config file in opencv-devel package
* Add opencv-includedir.patch - Avoid use of ®/™ signs in specfiles as per guidelines. - Disable LTO on ppc64le for now, as it fails to build when enabled
(boo#1146096). - Increase the disk space needed to build opencv. - Update to 4.1.1
* DNN module:
* 3D convolution networks initial support
* A lot of improvements for ONNX and TenforFlow importers
* Performance improvements
* Added IPPE method for planar pose estimation in solvePnP
* Added solvePnPRefineLM and solvePnPRefineVVS
* Security fixes: CVE-2019-14491 (boo#1144352), CVE-2019-14492
(boo#1144348).
- Check https://github.com/opencv/opencv/wiki/ChangeLog#version411 for the
complete list of changes.
- Drop fix_processor_detection_for_32bit_on_64bit.patch. Fixed upstream
- Drop 0001-Handle-absolute-OPENCV_INCLUDE_INSTALL_PATH-correctl.patch
Fixed upstream
- Refresh 0001-Do-not-include-glx.h-when-using-GLES.patch and
opencv-build-compare.patch. - Update to version 4.1.0
* DNN module:
+ Reduced peak memory consumption for some models up to 30%.
+ Inference Engine
- Inference Engine 2018R3 is now a minimal supported version of IE.
- Myriad X (Intel® Neural Compute Stick 2) is now supported and tested.
- Automatic IR network reshaping for different inputs.
- Improved samples to work with models from OpenVINO Open Model Zoo
+ New networks from TensorFlow Object Detection API: Faster-RCNNs, SSDs
and Mask-RCNN with dilated convolutions, FPN SSD
* Performance improvements:
+ More optimization using AVX2 instruction set.
+ Automatic runtime dispatching is available for large set of functions
from core and imgproc modules.
* Other improvements:
+ Matplotlib Perceptually Uniform Sequential colormaps
+ Add keypoints matching visualization for real-time pose estimation tutorial
+ Add Hand-Eye calibration methods
+ Java: improved support for multidimensional arrays (Mat)
+ Dynamically loaded videoio backends (FFmpeg, GStreamer)
+ opencv_contrib: Robust local optical flow (RLOF) implementations
+ opencv_contrib: Implementation of Quasi Dense Stereo algorithm
+ opencv_contrib: New module: Image Quality Analysis (IQA) API
+ opencv_contrib: BRISQUE No Reference Image Quality Assessment (IQA) API
Check https://github.com/opencv/opencv/wiki/ChangeLog#version410
- Update to version 4.0.0
* A lot of C API from OpenCV 1.x has been removed. The affected modules are
objdetect, photo, video, videoio, imgcodecs, calib3d.
* Persistence (storing and loading structured data to/from XML, YAML or JSON)
in the core module has been completely reimplemented.
* OpenCV is now C++11 library and requires C++11-compliant compiler.
Thanks to the extended C++11 standard library, we could get rid of hand-crafted
cv::String and cv::Ptr. Now cv::String == std::string and cv::Ptr is a thin
wrapper on top of std::shared_ptr. Also, on Linux/BSD for cv::parallel_for_
we now use std::thread's instead of pthreads.
* DNN improvements
* Completely new module opencv_gapi has been added. It is the engine for very
efficient image processing, based on lazy evaluation and on-fly construction.
* Performance improvements
A few hundreds of basic kernels in OpenCV have been rewritten using so-called
"wide universal intrinsics". Those intrinsics map to SSE2, SSE4, AVX2, NEON or
VSX intrinsics, depending on the target platform and the compile flags.
* QR code detector and decoder have been added to opencv/objdetect module.
* The popular Kinect Fusion algorithm has been implemented, optimized for CPU and
GPU (OpenCL), and integrated into opencv_contrib/rgbd module.
* Very efficient and yet high-quality DIS dense optical flow algorithm has been
moved from opencv_contrib to opencv, video module. See the example.
* The slower TV L1 optical flow algorithm has been moved to opencv_contrib.
Check https://github.com/opencv/opencv/wiki/ChangeLog#version400
- Drop obsolete opencv-lib_suffix.patch
- Add 0001-Handle-absolute-OPENCV_INCLUDE_INSTALL_PATH-correctl.patch
- As this is a major version upgrade, the old 3.4.x package is still
available as opencv3 - Update to 3.4.3
* Compatibility fixes with python 3.7
* Added a new computational target DNN_TARGET_OPENCL_FP16
* Extended support of Intel's Inference Engine backend
* Enabled import of Intel's OpenVINO pre-trained networks from
intermediate representation (IR).
* tutorials improvements
Check https://github.com/opencv/opencv/wiki/ChangeLog#version343
for the complete changelog.
- Drop fix-build-i386-nosse.patch, build-workaround-issues-with-c.patch
(fixed upstream)
- Refresh patches - Add patch to fix use of headers from C:
* build-workaround-issues-with-c.patch - Update to 3.4.1:
* Added support for quantized TensorFlow networks
* OpenCV is now able to use Intel DL inference engine as DNN
acceleration backend
* Added AVX-512 acceleration to the performance-critical kernels
* Fix cmake mapping of RelWithDebInfo (boo#1154091).
* For more information, read
https://github.com/opencv/opencv/wiki/ChangeLog#version341
- Update contrib modules to 3.4.1:
* No changelog available
- Change mechanism the contrib modules are built
- Include LICENSE of contrib tarball as well
- Build with python3 on >= 15
- Add patch to fix build on i386 without SSE:
* fix-build-i386-nosse.patch
- Refresh patches:
* fix_processor_detection_for_32bit_on_64bit.patch
* opencv-build-compare.patch
- Mention all libs explicitly
- Rebase 3.4.0 update from i@marguerite.su
- update to 3.4.0
* Added faster R-CNN support
* Javascript bindings have been extended to
cover DNN module
* DNN has been further accelerated for iGPU
using OpenCL
* On-disk caching of precompiled OpenCL
kernels has been finally implemented
* possible to load and run pre-compiled
OpenCL kernels via T-API
* Bit-exact 8-bit and 16-bit resize has been
implemented (currently supported only
bilinear interpolation)
- update face module to 3.4.0
- add opencv-lib_suffix.patch, remove LIB_SUFFIX
from OPENCV_LIB_INSTALL_PATH, as CMAKE_INSTALL
_LIBDIR is arch dependent. - Add option to build without openblas - Add conditionals for python2 and python3 to allow us enabling
only desired python variants when needed
- Do not depend on sphinx as py2 and py3 seem to collide there - Readd opencv-gles.patch, it is *not* included upstream; otherwise
build breaks on all GLES Qt5 platforms (armv6l, armv7l, aarch64)
- add fix_processor_detection_for_32bit_on_64bit.patch
- Correctly set optimizations and dynamic dispatch on ARM, use
OpenCV 3.3 syntax on x86. - Update licensing information - change requires of python-numpy-devel to build in Leap and
to not break factory in future - fix build error/unresolvable for Leap 42.2 and 42.3 - Update to version 3.3.1:
* Lots of various bugfixes
- Update source url - Rename python subpackage to python2
- Do not explicitly require python-base for python subpackages - Update to 3.3
- Dropped obsolete patches
* opencv-gcc6-fix-pch-support-PR8345.patch
* opencv-gles.patch
- Updated opencv-build-compare.patch - Add 0001-Do-not-include-glx.h-when-using-GLES.patch
Fix build for 32bit ARM, including both GLES and desktop GL headers
causes incompatible pointer type errors - Add conditional for the qt5/qt4 integration
* This is used only for gui tools, library is not affected
- Add provides/obsoletes for the qt5 packages to allow migration
- Drop patch opencv-qt5-sobump.diff
* Used only by the obsoleted qt5 variant - Cleanup a bit with spec-cleaner
- Use %cmake macros
- Remove the conditions that are not really needed
- Add tests conditional disabled by default
* Many tests fail and there are missing testdata
- Switch to pkgconfig style dependencies - Update to OpenCV 3.2.0
- Results from 11 GSoC 2016 projects have been submitted to the library:
+ sinusoidal patterns for structured light and phase unwrapping module
[Ambroise Moreau (Delia Passalacqua)]
+ DIS optical flow (excellent dense optical flow algorithm that is both
significantly better and significantly faster than Farneback’s algorithm –
our baseline), and learning-based color constancy algorithms implementation
[Alexander Bokov (Maksim Shabunin)]
+ CNN based tracking algorithm (GOTURN) [Tyan Vladimir (Antonella Cascitelli)]
+ PCAFlow and Global Patch Collider algorithms implementation
[Vladislav Samsonov (Ethan Rublee)]
+ Multi-language OpenCV Tutorials in Python, C++ and Java
[João Cartucho (Vincent Rabaud)]
+ New camera model and parallel processing for stitching pipeline
[Jiri Horner (Bo Li)]
+ Optimizations and improvements of dnn module
[Vitaliy Lyudvichenko (Anatoly Baksheev)]
+ Base64 and JSON support for file storage. Use names like
“myfilestorage.xml?base64” when writing file storage to store big chunks of
numerical data in base64-encoded form. [Iric Wu (Vadim Pisarevsky)]
+ tiny_dnn improvements and integration
[Edgar Riba (Manuele Tamburrano, Stefano Fabri)]
+ Quantization and semantic saliency detection with tiny_dnn
[Yida Wang (Manuele Tamburrano, Stefano Fabri)]
+ Word-spotting CNN based algorithm
[Anguelos Nicolaou (Lluis Gomez)]
- Contributions besides GSoC:
+ Greatly improved and accelerated dnn module in opencv_contrib:
- Many new layers, including deconvolution, LSTM etc.
- Support for semantic segmentation and SSD networks with samples.
- TensorFlow importer + sample that runs Inception net by Google.
+ More image formats and camera backends supported
+ Interactive camera calibration app
+ Multiple algorithms implemented in opencv_contrib
+ Supported latest OSes, including Ubuntu 16.04 LTS and OSX 10.12
+ Lot’s of optimizations for IA and ARM archs using parallelism, vector
instructions and new OpenCL kernels.
+ OpenCV now can use vendor-provided OpenVX and LAPACK/BLAS (including Intel MKL,
Apple’s Accelerate, OpenBLAS and Atlas) for acceleration
- Refreshed opencv-build-compare.patch
- Dropped upstream opencv-gcc5.patch
- Replace opencv-gcc6-disable-pch.patch with upstream patch
opencv-gcc6-fix-pch-support-PR8345.patch
- Enable TBB support (C++ threading library)
- Add dependency on openBLAS - Enable ffmpeg support unconditional - In case we build using GCC6 (or newer), add -mlra to CFLAGS to
workaround gcc bug
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71294. - Apply upstream patch opencv-gcc6-disable-pch.patch to disable
PCH for GCC6. - Test for python versions greater than or equal to the current
version. - Add python 3 support - Added opencv_contrib_face-3.1.0.tar.bz2
* This tarball is created to take only the face module from the
contrib package. The Face module is required by libkface, which
in its turn is required by digikam. - Added _constraints file to avoid random failures on small workers
(at least for builds on PMBS) - Update to OpenCV 3.1.0
- A lot of new functionality has been introduced during Google
Summer of Code 2015:
+ “Omnidirectional Cameras Calibration and Stereo 3D
Reconstruction” – opencv_contrib/ccalib module
(Baisheng Lai, Bo Li)
+ “Structure From Motion” – opencv_contrib/sfm module
(Edgar Riba, Vincent Rabaud)
+ “Improved Deformable Part-based Models” – opencv_contrib/dpm
module (Jiaolong Xu, Bence Magyar)
+ “Real-time Multi-object Tracking using Kernelized Correlation
Filter” – opencv_contrib/tracking module
(Laksono Kurnianggoro, Fernando J. Iglesias Garcia)
+ “Improved and expanded Scene Text Detection” –
opencv_contrib/text module (Lluis Gomez, Vadim Pisarevsky)
+ “Stereo correspondence improvements” – opencv_contrib/stereo
module (Mircea Paul Muresan, Sergei Nosov)
+ “Structured-Light System Calibration” –
opencv_contrib/structured_light (Roberta Ravanelli,
Delia Passalacqua, Stefano Fabri, Claudia Rapuano)
+ “Chessboard+ArUco for camera calibration” –
opencv_contrib/aruco (Sergio Garrido, Prasanna, Gary Bradski)
+ “Implementation of universal interface for deep neural
network frameworks” – opencv_contrib/dnn module
(Vitaliy Lyudvichenko, Anatoly Baksheev)
+ “Recent advances in edge-aware filtering, improved SGBM
stereo algorithm” – opencv/calib3d and opencv_contrib/ximgproc
(Alexander Bokov, Maksim Shabunin)
+ “Improved ICF detector, waldboost implementation” –
opencv_contrib/xobjdetect (Vlad Shakhuro, Alexander Bovyrin)
+ “Multi-target TLD tracking” – opencv_contrib/tracking module
(Vladimir Tyan, Antonella Cascitelli)
+ “3D pose estimation using CNNs” – opencv_contrib/cnn_3dobj
(Yida Wang, Manuele Tamburrano, Stefano Fabri)
- Many great contributions made by the community, such as:
+ Support for HDF5 format
+ New/Improved optical flow algorithms
+ Multiple new image processing algorithms for filtering,
segmentation and feature detection
+ Superpixel segmentation and much more
- IPPICV is now based on IPP 9.0.1, which should make OpenCV
even faster on modern Intel chips
- opencv_contrib modules can now be included into the
opencv2.framework for iOS
- Newest operating systems are supported: Windows 10 and
OSX 10.11 (Visual Studio 2015 and XCode 7.1.1)
- Interoperability between T-API and OpenCL, OpenGL, DirectX and
Video Acceleration API on Linux, as well as Android 5 camera.
- HAL (Hardware Acceleration Layer) module functionality has been
moved into corresponding basic modules; the HAL replacement
mechanism has been implemented along with the examples
- Removed improve-sphinx-search.diff, opencv-altivec-vector.patch,
opencv-pkgconfig.patch and opencv-samples.patch, fixed upstream.
- Fixed opencv-qt5-sobump.diff, opencv-build-compare.patch,
opencv-gcc5.patch and opencv-gles.patch.
- Version OpenCV 3.0.0
+ ~1500 patches, submitted as PR @ github. All our patches go
the same route.
+ opencv_contrib (http://github.com/itseez/opencv_contrib)
repository has been added. A lot of new functionality is there
already! opencv_contrib is only compatible with 3.0/master,
not 2.4. Clone the repository and use “cmake …
- D OPENCV_EXTRA_MODULES_PATH= …”
to build opencv and opencv_contrib together.
+ a subset of Intel IPP (IPPCV) is given to us and our users free
of charge, free of licensing fees, for commercial and
non-commerical use. It’s used by default in x86 and x64 builds
on Windows, Linux and Mac.
+ T-API (transparent API) has been introduced, this is transparent
GPU acceleration layer using OpenCL. It does not add any
compile-time or runtime dependency of OpenCL. When OpenCL is
available, it’s detected and used, but it can be disabled at
compile time or at runtime. It covers ~100 OpenCV functions.
This work has been done by contract and with generous support
from AMD and Intel companies.
+ ~40 OpenCV functions have been accelerated using NEON intrinsics
and because these are mostly basic functions, some higher-level
functions got accelerated as well.
+ There is also new OpenCV HAL layer that will simplifies creation
of NEON-optimized code and that should form a base for the
open-source and proprietary OpenCV accelerators.
+ The documentation is now in Doxygen: http://docs.opencv.org/master/
+ We cleaned up API of many high-level algorithms from features2d,
calib3d, objdetect etc. They now follow the uniform
“abstract interface – hidden implementation” pattern and make
extensive use of smart pointers (Ptr<>).
+ Greatly improved and extended Python & Java bindings (also,
see below on the Python bindings), newly introduced Matlab
bindings (still in alpha stage).
+ Improved Android support – now OpenCV Manager is in Java and
supports both 2.4 and 3.0.
+ Greatly improved WinRT support, including video capturing and
multi-threading capabilities. Thanks for Microsoft team for this!
+ Big thanks to Google who funded several successive GSoC programs
and let OpenCV in. The results of many successful GSoC 2013 and
2014 projects have been integrated in opencv 3.0 and
opencv_contrib (earlier results are also available in
OpenCV 2.4.x). We can name:
- text detection
- many computational photography algorithms (HDR, inpainting,
edge-aware filters, superpixels, …)
- tracking and optical flow algorithms
- new features, including line descriptors, KAZE/AKAZE
- general use optimization (hill climbing, linear programming)
- greatly improved Python support, including Python 3.0 support,
many new tutorials & samples on how to use OpenCV with Python.
- 2d shape matching module and 3d surface matching module
- RGB-D module
- VTK-based 3D visualization module
- etc.
+ Besides Google, we enjoyed (and hope that you will enjoy too)
many useful contributions from community, like:
- biologically inspired vision module
- DAISY features, LATCH descriptor, improved BRIEF
- image registration module
- etc. - Reduce build-compare noise
opencv-build-compare.patch - Remove BuildRequirement for python-sphinx in SLE12, since it's
not available there and it's not a mandatory requirement. - Reduce differences between two spec files - Use pkgconfig for ffmpeg BuildRequires - Update improve-sphinx-search.diff for new python-Sphinx(1.3.1)
* now that sphinx-build disallow executing without arguments and
give you "Insufficient arguments" error, use "sphinx-build -h"
instead
* the default usages output ie. sphinx-build(or --help) no longer
are standard error but standard output, drop OUTPUT_QUIET and
add OUTPUT_VARIABLE throws the output to SPHINX_OUTPUT as well - support gcc 5 (i.e. gcc versions without minor version):
opencv-gcc5.patch - Update to OpenCV 2.4.11 - can't find NEWS or Changelog
merely collecting bug fixes while 3.0 is in the making, 2.4.11
didn't even make it on their web page, it's only on download
server
- remove opencv-underlinking.patch as obsolete
- remove upstream patch bomb_commit_gstreamer-1x-support.patch
- commenting out opencv-pkgconfig.patch - possibly it requires a rebase,
but the problem it tries to solve is unclear - Add specific buildrequires for libpng15, so that we are
building against the system provided libpng. /sbin/ldconfig /sbin/ldconfig ibs-power9-11 1651987457 4.5.5-150400.1.28 4.5.5-150400.1.28 libopencv_videostab.so.4.5.5 libopencv_videostab.so.405 /usr/lib64/ -fmessage-length=0 -grecord-gcc-switches -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -g obs://build.suse.de/SUSE:SLE-15-SP4:GA/standard/02d6483556ad3bbd8c0ab597e65ba131-opencv cpio xz 5 ppc64le-suse-linux ELF 64-bit LSB shared object, 64-bit PowerPC or cisco 7500, version 1 (GNU/Linux), dynamically linked, BuildID[sha1]=e7db46bc36b605436089b299e304b1b84cac1393, stripped P R R R R R R R R R R R R R R
R R
R R R R R Ƅ|Mt (Z utf-8 4079b4e14f3e6ec8e743bafcef82c41fbbb8c8c587ea3fdc7e3ab5cd5b3c2829 ? 7zXZ
! t/㚛]
crv9w!AXWuUv P
)% K-[ ) TR,wm
2 1wv弝qs[yc?~N]E!EO v֏AH-
K m
:;Lg SX2g&&Ċ}V1s;,;F l`o#*l 6QWK Ly SZg
0v ;gxgPVB42NlogMM|uELIu Cj.pYti)X3,-ĶO22IfxMH}WڳT QL%: Ro#FC &;6qiq)Aq%EbnzE:ȅ](xr ] y*= "^hٹ".
参4]ےL]pdD3_^݁h$cy d&~\©8 ށl$^cD4u't(d;EJ3Bl ޝ)rhmEN:K1R8sG[J}=LКk8tJ\#ABWO$?Hj9L9 :\{8?B1tZNO1X: 楓ō6dLb7sUjuHcЕAg C E/%$c TGMnv}í! V 2"BR8
,:2sj&=mXY'k{gƞx˿ ?s") @SW{ K~] I 7 mqU{i [H4wz^ G ;Anv,ͣ頌
!LҜ &N_EBpgtp
2O2J=0ϭ1_d\)_J$W-P
&ѵqUcA0}&1{6 Z
F#\F>l4[6AYWxb1%P#ߌ513HjC>R9N]eSk517o>גgM
!RKPd\cQTmγ|t6ʼnet ۫+1#*F cwG*5"ίOiF哟#k]*Ġ3Z 7 òX^Ʉ88TL7V_/bbl PK ݄ɀ@OA"Z;0T?0TA|8j2_ g ,{s!qQޚ Jx5YqIGQԢ|FLz/'ؚ=iS%y A5fBG/ {1Fm3 GBg5!yc N3kTjU֪J#X=d \}=.0 &xw&j]21 x 41md zNܿ<{&8i>[
85]+YY6"2elu YBXFΊ /W;. p H+a0cIUGBZN2"=oOyL`"˖wnM C߂cݮ^}ZB܇LeXmJRnEic?Y YTbTZn6$T!`1pIsِ<*Q鱬g@wTq*s7]AXm(۵RFhbۣEK*8}o jt?l&v#Z8V;h۬\:#'/YcN2D92> W}͑H3|qYr\5x9hbфbW
n9q; ,ۖ'@b4w݃8t],PJ'*svHj˻w!"Ç%ކ:{v+_&/+`\ƙ[FB>V[ ^+4&?zGH/D/aDH~8X*BUaDŽ= p(0 kwsM\%.K;Or)Lr_ܿ]E=): 9 *B'so
eˬ 䐰 XUG&Foe r3 +AAS; Tj ml݈m7ƭ@ sHyp,XʞEf,y6dU,ř^aLh7PoWHG"M3qoBVC1AR|pi*"A&.k͑={~uIP#Biy`Q)ut4A>9 eȝ/7O ,g\Ӵn ?
5pɢ]C,n]%_~!a 0A)Oo !D,}ON`$ZnG SL(Or,6!`8Vq/> uªo
Z>(l/.<7 llNv P3T( |'KէMS㯚KlҙfX9"~-$3 QåB+~TGO736ea]! Xqt{
#L;$.AmsL;(\g<O %]1b`uyCܘ4M Qòc/
Iix e+CT
6)! صb~@wۿ9^ZBB? [Tܜ/k%=`;xӑ$:i#m9n.IDsbܗQvzӉ ! '`I͡.Z^R&:xٰ:} 'kyPU`UvR&Ɖ v\Dqo?fQ1nF84Ჴz(7!AA~S)G4q@)ޡ%3aBY0 v{9,TH'Tq` !fP
:UA=X+`dQ+:̩eT}4㬇&YuG9[ۛ@;Y5.x +,bzMk_(Q7mq[ө(Ea`$ wϽ e @PצJG>{b♝F*.Y)ueJObqb -gX"\AH
v1 E%
<#XāPRܣKo[ǀ bT3xmkdBx$2oxGϐDB&oGo `c5(Ym O*֦~ ğtx\\Mcp Dƫ ~[ ɉ{j%#SIfJTW`KWEslྨ{bg(zƁjp10뭱RjBj&ČsY,j#̧+k68z9'~E̾ q %;rwWL
Iq>b ߛV=da& GyXuYHXMClfd9h0 JɆ7 a>!E \qu7On<_36Z"MKtsf)|U I*3Q>ghb7a̜WlF <[
`7 &>a75Ub7S4S +VTZWsRY^txog6>_ sNѸ1q7!Mp%rh[{8@&}#@_&)G':DuknLvU)ԱἹcLaC1B )E/71!\a; F3h(W2\@AɇV6խnԛ
^bde^ Jua3Pf c+RbKN/ɱp[~%3ܼ[b5^PzĂ-^0#( W*R" QZ&E_ [x=#Зq eU&>=SBd
ZWUQt}YB'd/k4Aj_e#m,z?*v2~+ֳ;D- qi r
{Tib"9 : CiNx"+Ǔ 0
ly)$wKa ⨥ TjL%I-_@ G8'(7PqP^8/S"o>T۱"*Fl?$-|GUƁ,OKc%GRhEVWHi* xe]AyE76mI<%D'E5R=ZSjr$ϲ.SM s *FU2DnWbokа7$ >_6@|* B`nyu
Avz91*+ RDE)6AE8)Nbk.0(z)L:Xr)} NP$9{8dE%Dˀ| N"8@- ^9WbQDOcFG9q:\QN-Mm_ G@|tY`zoX
T@2rvH&ex|q5:s1aTM5%1@A %7
a ۄe&9 Ѿ
n%ئ6l)Uy ]wی]LA(
$Gox3iKl
c Zehp
K֪ʪM)cN SkݒvzD1/?I/u l=n-¹_Ns!ny䈾*f;5Uf
s*7Qވ{h H ,ʵ_Ύfb82!BEΝx1
L }W>0
-p+ezݰYp 3 zjVQOt7G.K:@%*;ZV'_oMqݦ:JȂ.P<?t?
J(D'G4!" 6¬Ij
#y>X[Y =
1dS3gu&(:pU=9qKu)6Krxq*ؽD?.P$b7, }ݢ֣wfU0]aJqJXM:XOr⽓akNԡ}v +uE0v%%w|@J:ݿZ^~!|4ã$?Ӵt.[M KKd&[n3VB:7`n-c̨Mt,aP/Op H]s{>qaKϒ zޏ5_t33TD)MLc8O
)D46 ǧN=כ afi(L dVlɪJ$N!i@e\}ɵ!j͟
B^.yEScU3}ݨocN7OP1^x2NvkdQ[#aҢ :pMړsIc fu[kvO?,Yaf #lb\ o)%F簢eQnh" (]xPkP$3B
|x:iu0}DyQ̞ u3D5ųgq=fЎՄq (F-q ]A/ {f #A ^&,q&qX:j6a901U&Ȝ|q.k;cGͬQ-W/gdfb16D'@d. @ N0M( ?/~mDa
F'7ܗ[ïyS߳an+&U
0 Ha/xH)pH᧗|0 n ۱2 tX(c#T
v<6G0#0Lq(_bP,MblJfݓjIfk1g$a ' & eqj{ 0| c8,q8X>&+ry.wԼ%;ʐY>$KS0>
5@Z(7kCMsٽCsCV2N;q^2bKJ7F=@EoY dzx${ SRE+F jk=F<1 lH0@0;wO3+ ̺@f5Nk 0mr4ل#\yL;Umrȝo 6|bf/PE lBkȤb~͟v"G⻃.D=CjSJ N7O4_y5ތ
JfhKs Cf/aY't;@
`ФkｧOs W1BPt,O`{+D%6hҿ9ڟJ;h?^Κ䎦/I i|6ţ,):g.~Gf}[D9{tEa12^֬{' |wDΒL%PjqfP[+WAYsews\5@/pk|lu\u՛?P0O~*NFL ԤoFuCi4k&@_3yaf0'f4ftpjl 33CZL.D &!"op;Ò);mt2jbN_kN gYǔ{"سSr |&)7/q^_ \a-Xmݗx+K,K]="vYZژ#Ѡn vcc[+`_˦ 0n-M$ktoY*;
WKU=g.˝GA&ka֘
K#4nꏃ:e DN ~( O>I[ !{ߢ[IqtgwGum_T2Q(, Ⱦ{>?V"
A6:Uu"@uJVxr>u~Z:\mƿQIg
Ӌ,|W
'E
tU(Z;6_hߒaW .W2}0 cA?[$ȧVƗ jG$=PGwIb;ө%Q&S~X~2ۜdk _ҕL\פ
6kC65O=A6ĸBF+ :@ÿь$|/-헲3:GgZ8, jKhPD@KH)Xd&} wy!zϜ K~|s:&!5^T 跿9)_zF ė8)=DLW3-?II6&L:Smv/_*;U 6; ^ufPAI-ЈYE7 W};yˤGuvz(Di ,BM'IH2SbJ=Ans}CR2pPbNQE
@_H)ئzT
|]:@8l
;[V5:or+ff/d&`l9`!\Ɖ"b!B@ #4y du G;~ ##] w`ly ,MFAG>ѠdjlʘL|yS0(%Atlf.n8\̕i>/Pߩޮ> a~6TW:]*Pu&[{'ʟBָ6
Hw]:LcZb1?cI]Wwy^ /)
Rͳ +V ycAhfRy^採^>LV .}J[? IME.s,Wpn qzx_1۷$l8թ@Ubh@-R5MA%T*r&ߠ@O *.qF
0MDw nQ`M%Vs2]7X7ʀ/ RFO\<> W'uaa')oTxv0ƕ7i: ?>-J
P:xx J*`
Ɩ1ҁDǫsr; +̪Hn1g 0a1;14x<|qgw#] xhdv w
}p](
`#u˲5ӤHs ̳Y&ᕋv \,^p
."-c-`4F2+
<:ه64 U6,Y+ݱp.d`ICqPe Vq`sǃt'
Fc v5Pq7;ᑵ IB(;zT*bN5Oͮ*J&\!$Ht=U]ˢz
BTZ1rW1M
O59\e.7g!Rrߜ?K+\^[[Z7g -9$0T$N{ @3Uþ M`[O٭6܉ |pq'z u ^)lS-_ȃ݇M\nz`3')'ʩ\"|q'B]N Wd3XP"HA !?ڟ_HFיZ2:-D Ήξ-@p d gv9_vlI0' z0W3Yծn5t5~\867T Ès~4 u5*+7HF9{ f yk]
@RvsNʼ$$|u~ s48$SfkN4Do1*$w74~kd:dJok)6Qj{ Ӷ6[ \}}wCӌe]{ fA&{:aGD¯y2,uI,=%T;X[`.X+gUvܕ6db{ܿ}N 9LT6ݓ fN {跚8pMd2 @ݱQrڃ `ؐ1_o2F&&D^21
Фx^e. 昣Gk ?lm^Wmḣф&E"B?R9 o.Q4\M!wG-S {o ,,l
Ud^5@ݪ4pFZuDin7S+'Ȅp&G@ftT@-. $N͆Jwr MW+PVE| ~e8e/+|4 U>l~q4t\<wx_5'_6^8y$Dd v:a DJoY?*%Ww?4gF.
I[k$5+mu'Y]Y0fV# 1T ${-XG|}[[t)VtT(`R1D
>0#*& ;埇o YG2DZ ^ o">ȗ C S23;jz'B?CBp(*PԖ +E:kP}uה (ۿ*=`]
W$2~+?WU ZWpu{2 1#] ;l H~ T\ -Mh4koߘ/falxZ{Y|nj( " <|**]X7k%XZ(A
?cЬݷ RH7Bsiߢ4ʒ=JPi}^뻫ep c}jBzS#p 66tleQ聑8E;f`ou7kH:'Dq* c?MTwqZ ZRm Y# b '`GgbwSAo m&/^Ü*Qȥ)Nk~cL m'Sw;8́{l.$1ؙ;]=#jfT L[ 4˷QsC 4\anX \MLEvz[ˊIJE&lRS9J^}R `0]ry 9 ݽ x[ZYp9j < [T -'6.v4b.tg`'$?r՝W^溶9[EiƾUd
m
/RHkc 6tJ @|/0ыOwQ
S(f] D@VډnYiEw͙}YF?*Wƹ@. DjXG+0ϫ REν;:GޔmiL% ٦ς1ܩ`vziB IYEp;tއ*Xp'uVtߏDa/n vD(r,a "ˁ!ѕPm"l6
jZo
k1
, +-3S*]CG>k Jyѷ۠`ѩ\E4ͰiYhYN=7r#lǦ!amr-²«^ulce,Mb[UJm-tAt gcz[
g]Lt9yf)l>m[^͊˱:~\hga5i.,=% s_x3.9Õ%M#{PQw}6RB+-+ʌ]`߱)F1 Ӫ\wb X+`TӿmCn]"mc%n=%l& oI͟OJ_fgZϜ܌Xh
%0B6@bFGaQ@xR\<#8]k'0Cw#Pw ^&0]uB/核cu=٬g=p6L_kp)ͳ(E*!kql]q9 GU߀F}O!U b ԍ5n22oTz!!fcVG )_!v[ 3LAstD?g `
*(xG9GjԴ
P}'Fܓ6Y]d[0C ²b cH gi>twRrԶY7!pep-?2[=Dob `l TːS USV|@hB/b@F-Ă e+y $7C؇nD+4De94&wdӞ0Aid' ǣ ۫/d Kv,na>WЧVwBuGed1`z{l K+GNAV'qMve? K ̸hGwy=qGuA${Rڂu Ϭ)Ec@S\`hy0 n֮D~D Ao)t;ݖ!L\M$#o[ ߽2<4bԀֿșb7v~y!i5 ,`WF\N/$"w8:SJYK& Vu 863rcQՀ3y\B$i =I= Ͱ\\ ⧒MV>Znܾh
3Dh.hpϸ&8ܧttp($ L}3Sr[jSP,&-
T/7eVU)40_OIN
(McEGeŁZ(ix
=׳O6!L%/হ!X4^Ƽ<$0 HB n0Ogz- y%~ Si ZEs3rbNzE"h6Ҩ3)+Ta7BhUESJۢ[h -+ aD̡7o8 ny\XZX>e8QeũYxP 8hxw
,[skޅZ ݴlKD4Cp`
TUw B|_v;V .ѭp
<|D}՞#YX AÏ!r(2*1VZ>LLCJ/@ և`>jx?R\\@Mmp_ TJ+ H4rgo mQT3=yRn
3C# ɷM&2|lj :J#.1D=0瞂麯Gr.A'A_ K.FV4MʬU
+^.#Y /"[
`1M9xatDQ+ (CGG|rX#e-8=:ΖyZxQ` oC̝|NG8 T,2n5ܕ56|,mO$ Rb}݅=gEZEʮ.Wލ7Z6ǁ
)Zs6G8 [t"m)&5da0Z<8ÿBWMKF
"'!hϘ#QeH̺ F 攮;;Z!!M3g=0Z SDɼWM3Dxƚ+ TaxwfG -p,dB 4ٙ:W`pjp7ᆔ;I%2>j2|y.ΔNBfjO{9oX` O {hd?Z^z\wsh u};,۶1d1]s5FeDf_m@zy+] %9QM{Tا75(pS 28 qNhaRR8~,LePi&@Ҳx}:>MI?B0ؐ>b{Zϵ^<cL[rѸRtK * ?h]`?ǑGz5
xшSJ8nq6 i!P Kp%S~H(#fJ8,xyZJ 2Nv>"KU@^xYvcهaB VvX-ha /cnTLΚ}?
GHjPX
^ȂK(⃡K 0D'kBF|7 bGE._+sUxcqŚ8"8op}A}h^gё8wY,C',
EǠ o([Rap~*rReϧTF 8WQ{0dЧ]q6M"ϣVS`uN#Zđ%rb-B^EIn2#qx9o6n|> Q
kmGgdHbF[n *dm
0B_
LCfgwsKސdP]D@5i ZԆsyz/.vDP 1'`i `"w * 2hHNƜ"6-7
`z0i">҉\ DdZ߸aP^-)5C'J%
S~BfY}AӮƽ=CgZJ, Ikɔ3# U(e:#Rp3̥9RA,sem&@Y~萒Nq*◠QL
'!
X}|hg=U-!ѐLX:?xrz#rZrˑTicGԔc Нa2M|hZGZz Kk0%tDUA90;9Fp@tɔ vG ]W> fI`y '}hi6K0yy+fe1Nkz5J+ȳc @ H`CVkjs|Q}Wxv]m 廌};=j$35=p9X)|R#̖04*To om`ZMi !d3=1\UZ4ȱ֏ a=}n5^/3ߒBV)b3!_q1`)Ϯ+w5
)R/i^3'#.,)۰<>3O9(J žS@J<gx~4t S <)~~691\pp@6
?D36X7Ik 8VQP xL]1CAHm. #!r Uu{k竑\G2} ˳'y#N ]8lO.,h+bVF-ۛyM# -i$2yHb2Tlv.7A
7U=9" b]E$!վojieM0PRG_?mIɝKAj_*61<~Ġ$@ڔ wGMr %RD ciYG
8?= 0I@HGX7o#a$G mdUDV3fZ`@u;9od6
PE@7Nȗ[wDi[*i}mѻxK|ݍ Cod8tXw) KCoiHќ{sp`}Js| 0)%҇B{ JQ)T4H駓KҩCĐXz,dC8_K0f͘ d\UÆG+6)M3Ո~Q?YH"i2ݣF^UklY;vfy~Q$. *Dyx"4jY;` 4XKh):;LV]zI>RSWbNl]r j EPv8]EG's闕} YN`Zk0A3=ᥠ~,$sU M ]PnkIM$'[ț]R Mg pObr^2#<]
WbOr=}Y~^LLF~DC4xE砚%wǸ 0u:bFe~[²plI
Ʈbc
S8L-~eւՕ`~:b7 zuZ6LW *JW 5_
T ȇ
h+r6LCe&!"YfZBavts*:@xMCsA>$lb. erjMZ
y>[:{E_?Iz)0ێ>~a Eo< Dl00f #V+YT(aؕO^Б1e #\uĄ +,(U+VQ
.$Z,=U ( n療U {1ȪÝ o(S ]K{?nEAƛ\S5bPO6g51&I6|N2DY;{R;z8ݺbeQWBʊWBJRZZ9QdBR3̶oܔ =n nrH"2:]2 mI0# ~oR(&ABa0pr14e ij0
6ްMU;co}(qx[CO]PFcUj
M^W `%VAovy.56wEӡg#bnЮ "r`g9~
^IA>U^ E:K O +:! +
2ZJf.JiöLA
jl i4[$UMz'7'Й˖ѻ$7+°K|nF@5\y "rOgTCv=Ӟ955?
U\"ӎ✙+7
|0wP
߄rXMW05IK }Lo??<Mj4Rvw)`&lXe6vMO@sz%ѝ41^II1:ޠp)!0mbW
ZX6~>ԡ$罻H ]uN4xJ4:7 =|٪~{ٞ~ZqfwNC9EZ +AL
7B( 0e%*čۓ־,1'Ġڏ:8J!` >m| G=GZ">P"⁆pg |e\Qr(5Lׁ2 O֮&'),̜0,8}ʀQs[-pޜL$ta7~&Y&-99 ;edNQ3PR$IdJjzMmu/cܙ;3k vi;HMheNe它
^eL?y̥Zp7 AVp#
#=a0=gЧIy Y5 V%Hpť^9wN#_+HcG~lѮs/ ÍC]q
]k369*0tߙzVOj ~ : zv!+Ήb
/#c`欇arE 8
|( 55hLa.'FSתJ>g䮶eGQƶ)R3/~R~jGydih Pv,@dKV2%dQ&6{^ Й[zңqۡ&e(3 -AHMihp*$@5 {ĐbqTr
u#I6QsrI
]!_wWVa#يYTc(__;o\V AX'cxr+=$2Mξ3r4k>oSvd
JQz[d~#~0AC' m&yeBJЬX[ D1oGR&SkiÖu"RGy:F~×_/cNw]yP
Mv-8@kb+Ov_-Nh6~:e
)g (Lu~I6r ΐoKSZ5 sH
BGI˞-$ỉod<X햁Htq hDϠoywf8L*$vM0"`1B%D$&dd2uR.OJ
'd=GTef=dU`Wͫ`v% \>@ a?Oمw
DI4j nێ39|;9ڂOnv a= H7?+~B hN²~= 00Z{V~U54 , L=Q^flD㝌 Ay6-R PV0pq o$0ٮbK Tld+bǣqH;N\ l8>x5YW\r=«zn@:OD=V6"pv嫿hv_r<*3maW,N>~v;/j!7ٞsg&y75J%npϘSʚ'YJPꔦ.Z]x]G
=)ЃNglgn; ԍ5q&6n"XYz{x i`IB&Ky arAgw
5fN;?dct&}W זj~; ) ta
TH ,/6ИjCo^ެϢ*Aa(qz,E`Tsj9Eͨ
Ԁ 7U>p7lk _31 Ryh\i;K.*Ҹg0Η*85aZ:R/QI{؉?я(YS)C~[Ui KGP+D6iKc+j-3s4OoFyǞ=Rte{\_ՒvW@K7Og4tOxs{-^=נ磈In
{k=ŢCxv~y/n.
UN(n&oe_~CQ.&μoEkW_vz67Z8~.@-yaіK>ǹmxfp2M{~|-G^f&NI -*L۶eϮ- &ɘr?q0sA
(CBd#A/WcWBr4Hr
<aW:FE6&]BDϞEzE@a0/_n pe/Z8恅YcխW[*ڭz *~JlcY؎Mx }ǵߓwM!}J}BYqNSTp)oJG "~RXGFޚ]![qKQ#S>9iC?OKe[co˸HuḶXd̿Ox
{acu hԶ`YTCOnNf`a/LWoJu-2;H Ȗj"Fo_4JD~@kQ-- "8)Z6>90 !/e<JFNPHU1{jU?va&:$* `gG)e] x}6emY>AOX6b1 w^36 V|VPL HWծ(Lt#\AV^ID ?
KW<oHwg9dBeo?*LMaD Aə24\ &/ߙApspP'*xa +Q``bT4߆m~$+^ 7aۆT~T`0M۷Y? aQHѬZq n3蛜Wj_ n,mBJj"+Qx/!/s %EDol A0B2x E1
$|zv ToC2t=ͪ$;hE ͪ +&ȐY r' ]K%6ptəZ2+ 9:RGň~sKqGbPB/wp!J+xqlHG)ذl'pc[}ki$v]l{6, x&`;k<(
o~Z9Ob r))9sCbddJ{P @JOąX! |, 4%]\z9GF <Q3\ÔV9
^YL+Z7<8u|@`x58'lрRlx^Wx5 `RwptY% b `EC٪ 7/(]]叞E \k5f~^07٠>ΡnBZ QXkh2ϳ@ kkk<16 ߜ3fYd'Oʦ&$}UݚpA60f`g$fKQ83i~. n6Z$E]D=o~txش` k5|Mh$Zk\
dh$Ej H±\" < UMCE7c,X͆
8'peYҷ/b$ +#ڃ~7 _s$w('{+3d!i yO4 Uz=vWʢȐs,^V 'ök\d^w%yZCrawX~QeoA?:2&3 i<
p/T#rxѡ"xǃ m
qhQ4[MJ47i+lG9p#+S[u`86>ɽS= ?v(uNKf4WRϳG2,}D8/j*S\S2r70Z"wEX 'UgWz.FN.I/`?K߂f bg|DH9N aꆦS+( 4~o%- C`'Cxv*J5q7hkX`
Xc
!PGaE4j~8T%5@I_/>sʼxֹ`a6][?a1uēVH f] "&4iORi`Ohi mG_x~쩨Heh C zdJ-p4@z03.7qNo)зHz
n0IoEߏ|؟dx;-Dž""g]I NfOr+
eݜ
rac'i !N\~`|: Sos/wLoo.DaC!.suB?[[ R QKPb+8)ri(k)h`T{="3VI\@5wZukA}ˌ0LeݣWKa]< d[*eF e$HԎA~<_cm te3HZNUP14oaFM.rm8[ {
{Fj6._Mz19foi# ]^_Ԩ J j4Hso"pS0BޞM7 fYt !!9foWUድM֗b 'yH,7vlS6vm6 c5B6nl|kJHp
6c'A̳KwmL2m {Qsv0G`iB#ܰOUJCj*S( U XVņ[ CGj!R.w~O!䜭焙*Y/: ·HPpȉ (hҤbΩi B-~abΜredC
dY܍\F`@q }8~}.yк&l&U8t#hUpcOx
PD
ARo jݷXJ^* u÷ԇeI0:.%oU0۔7-㜈< HIq-h/zͽ` ?E`K "s&Wy`;ydz>EuHyg Pu!U2PiH ..PEm%eUp2׀˽- *$Cm HN:, {V,\E kJQDl/pp%zP, ["@p , aLx@ܡGq-Z{p?EaYqZ:o͊ѭ(Q5 RV9[mܤ\g9DՒTvr@{ʻ^6w`2L^9<
ݫ ėh\' faqG(DO%D|HH( <_5D,b"\tgrdbq_=R"z m,r W.8
"=Lrh{Gϊa[͡ ܠ
j4f 0_ 'DV*?G Rmv pj60Q%
LEp^T =U
o6#8舰s ' @Gs ByjӴܞßZA)_6X] s}`ͅ$O'5ѵ2UI,@o6isqm2g2ɯI)@irR]4:w hG ĔFf~;EzQ >K-:qHno_R^7Ұ8 cHb+"ZC5ݧ%)J5>V'{pns⎼ݔ$ |L&y i
vK#\{]+Mvdm]?fJ ݒ,j.b?-o\1x҄@Ɛ x Rlog`?sRc*蛎gҨ}P#K"M aAQ\Svt,]P`8ZZuJ rj-l Bm}j։OHۣSFL38QYNYɗ^LٺO[^М z1M*@ޕ,1ed.t@\zڟn$G7M(T`+ǗֶwhN3k@S;&%^<ȜCy3\5: QO rd7٫:;j9 3ehk6Ls7)Ia˧˖m`|î\SX3$[\%BnɟȽT LNy7z:!+,GdCOXm$ă ;Y{e
Z<`W,_Qz"Wwb~À%0 G1}OQִN-ǧਞ{4yt W4Bؤ71JQ0-pZy\ <1ߖ:UdS%c ٖFoׅ:(NY qwld(`= ~dp
l^4pB!M~_Ijg".k1'Cl,MMoUCYV/dw7d
Ȧ7up kZ('39]nVJPǔ[Z_ȩ"8je 5E)=C}_kĐt-P&y[U# S/lB|=d2Dh!P!ۉJ(>}[)G.pM+l 4xGh^v<ܩ{=DN\8CpXfUZͮش4/0-c[Dv< ?\$Oˁ> ~暯e0E"GC8j[(n4lOW&B c\Pq`F*g?|zBpp Z\[vSjxccq|BY$jN4R{G+-{AkA/sa 1`[96W XLeFpFn=~ e' 9$yD
9bMqkm~)F4։>[@N[n8u_xq5,ߙL֊ܼLh|ds20RH+࣐'&Q5zeE.o PU Uc좣ѽORPOp*-clZɾOLD5X*9XcPWZ/h`MM0~qvhBҹ5KӐM*IeSpEhKC|,kD|EjRMm W J镛Z+@+
Z_
7KB
_h5TLTÝru
?)y),THW ﭶc
d%x= iǮՓ|, Cj61e
Sv h `dh O"{Y UO25+*.Y42َW:{~mE^C¶- `ۇY 2 50 $ ѓ݊zG̭2&cG[G|flτ%]?+?j퇒6 `o` 7GR;/FzpYje0jK}0[̶Rt5d\N^H++d|R Ïڢ zz|, `VYqo^i_L5@xv2T^TY QvCI& "?(
Hy6 tuqk~: ע! 8ɳqNeIİ ٯ/7bhG t^`͛ؾdS ưF[yBa 1JW`XE'sl p RY=t9N,LlՆ LʅOr :8'#ۇ
j}<" ~߀<musu<, V!4.r{ԅsM[tcÛVrD-r7H*@Ik UZ(qD;<+n@}W26Met_
|ᱱ&6 [ `43.t1&D @/fgfy-PDb6 5^+>.E h:Hw`s$rXCfĀ6&DO96gBw?tGDa'Tc_ Z
(虖'9ttt@l!.
䇘q[j(eࣛ0ϣ y4־2[R# ƾV~ E
T9VҤ &T/WӉU##1FyJM#9ـX:l3 0_hmQ|}dn2~*BwNkkX l,쟞4yETn.{]bF02T ӬFNuZ gxuI, Z|͞ !Nt#:V9:3RC #bJeA' 8{1ˮ rGMÇB۵6
<2
mϜ\++s҈
Ҋ:; j9P.:Q VyQEVHhPT7KD
.~'N ̢`{{Z}ESc[7/ m 9y%|I({o x-d`!{*M^ k|*x*SXХFl)T-=яTD6n ֢^$'J4 ʙTjr
dɖ x;ig 9m;u# BkJȘϪbMj6d &#; :L: ܞ]hil:>WH EɡցuLyZ|NWnvԃO>* 䦻Fiڎm>4b,'ObԼ^Mt{Orc.pT;09,q5yO y2U1UNh|v?o'}o2xf?ccHIeI`"GRpˆj~Wf ҾDE"u> )d[,hӝ7gU ӱ|R!羙S +DDj#Ma̦wQ(E i?9ַXa 5$gim3mAe\v!/ЃY{v8(h8s7"M$w{RvlPDx _: Rs a=L˕+k7Wp)TlrqO$O"p7rŗToYnԗ\H0phAU ϓ =ǗXe FX `
{7 FRfm
nX%m+O>=Mқ_=- aڡ^X->ݪuIdU* )xjNg ߿w+IO*ԋ%qR ɬ5VR_ɡJ-\x`X2egz /:ol`ݺ;6!@_.C_?ng֙ܔaPET'& s
ٿ2d Ĵ 5 8)iy0;T{8ޭ4x%' Nɣ E`VNϻ xN0 d2JÝLew۬܊lhCFݝ)Be线R|, Lz]L\R!r T 7 ܵfmS~8 ~(}y==_{h$(dC87gM$ɨoOl[4٪_ +DAܿ.WqV xQUrmjxQ'8$yKdAtlOf| fm vB]rt1NFG+Y$ۭܦzqŊ)'eYRܽXh1|[HQL =$~b;CA:i i^NxI &]^~eoLac 2`h.ܺLA+! OH gIHDaҪu=Mc~}w@ /
[0zC.|Ə$[dNsʋv sa& Y>As^u6=Mt o$%R,eB_upm@MԗRޗX |/,fg݆cb={1Yq-lPEP>@uYҸ 7;v$iX?';:~e9_[?ڻ1n{(^ =+ީRxRpH'WO6Gથ7OJVT*~. .ٝa/2/|oҋ\ɍڀ*O!|]}Xn(' xb9R7Uf-sE.^1uY(-l5 500IF;Y ) @ͫ# AҎ k
+1ۡ
츰%#TU\(1` Vtc4pƳR{L/&YÉ,O0Hi*&0\xL%㉐`?G!P Y *v _v=?
gn0<͒'u41 %K і^d,:f\TB:t_Γ[QS~hDb5Sl ZacwX/ja ?!EUcm-"98aͱ]N3nfOZ6F~"X
c:Hm%ۅvgS~-(9c6,Rm,I sADR.Z$#K[ ګf8XxфIF2Oe [8^8 CU% #ΗQ)F
6_?鞭۴Ao09ʦ7qm6.iP^!/3.}6!SPd [HMpP+wݩ7+ˆX#(]B B(ϋl:"9rLma ;"sa:ia EehxsGKk
icSF +!49!Y+؋L Jf4lՏ'$be58rp-sN#dA*x}rS~;pvY-68nɢH!1c
zBEM l(6:Ba&P43m?#(]isG흖FOԼ wtX~VQ&k3%Nu?"tìFKc{H >OjrBdT$Mچtuwhؠpb>x zFjʼ ].> R14NB#JL zt~ϘZar'kc. ]<`#fWT ~^Ut~ Kߏ.zabXz35Y->S:
R\## P&u!3GUmkfۗ>@ڍ~Z@ٽ qRފpUk ]^=J/I,9YX)R O|}qf.l@v$!soLX[q֘m(ȓuf=ƭgpX :W*kVCY*qI-gb}UqʇGZ2 јGܨ3&@3];n^n Cr L?;IVQx- ~r`կ 3f~CGN09JsonvY!CQ=ѹ eT|F]4!Wne!-|ծsnˁ !m=NW=x 07!^H ^@c6] W#zG (OFءȈ vn|5 QZc AkVz/jev*i^mp{#oj1˶n%|yq!Bf1ўr@ֽAo~c8!T!0`wFxslX( $8W){0w9caHOҬ?TkL//& s_YqnMP*V\!naq3~k`A t JQ!FD3f ۮ2a䲙%97;)՟] R;x±h>:٤4}+4!-I;sM8 e3
Ѩ7FYw_sU 3O z Ekc
.ycHR3ŗγWeR:$ioCya%ld&Gv*ؗ2fkCУiĈAŀd#ͩFhFc;`רVKaHz +U֚2><ע V4w+
9C3.96Se!vfD/+8aNSu AMʑ &]R*p\?;7E ̈M!`QA0) j3mZ>rK3bu
C^Ip21#H.
\c>/ ZDɆ}O ^"Օw@LX!
Z͖)bN asg:Z|: 3uXŏbg BeqBA(O$=yj "?r;bV+m\:ԥ"X@q2HF0egNRJ^Eɣ~L_f h e`z2 >PYW3~D;/Fx 18Em|KZ,
B\0S=%I7K=Zԍ OFvO= u3s^JiN LncS 5̀ G)Xt,b
N랍 l$ 4vs$+Ƣ[~͉Ÿ VPv)_)ߤ#}ٍc
\O"~ $%{Wyd"<"!plLo9ݫʽk0_ȒQh\7C]Bm, _ iPSrs 'B
gADN,wĺ)~YXb$ ltmk:$d'Ζ?
}?4v$Rܝ2 SzrcIP 3W)ǹi@=sr&ɸ.ӽ##_("
6(o^RPlJG0M,n.SzE:W3 &Qxy-KNw! r[C =b\lSdهqơqj(iqyKR6Q\` S j .ߔK?|U %ZؖTC"Vxlk:۸jY%%7C(HQ5_1)k{$G`ŷM6 ivDueuJ6yzϸOȌ_ţ7n|XR¼1Ǽzm %)XΧ "m̟H
8v%e$bՖil'T7eHǞ0DGXb=oA kث|n>arjČvdwr#zݏy?u tP` IX[ 0 %Om01a>/{
d+mO{AJ<($6w ,7^A!dS[j|ЬK-P!}Sf&[k:fd3- ĤZ_y
(t gO1.
Ϲ Q"| [ZL%}
6Bʮ9{FN5gffaU~^`tmZ'hחUݸ
}幏Ci82@smGy#mBDɷ\?2Gv 91l^?ӼaV?0͵[fioPlFԺVcƍ%肒--ؾi;oKzS gE}xdW X~IUZ]"ؠ ni.W
H ^*NBҢQVd _]фNY IX۽',> )̙ "5 2#689g=, 5a7h-?t'E!(: OȴB 86/\ҤL#olOI'
s%U{b!aɽA"E|ϯKIVqZLlG31d5f/Ҵ{ b>`}/yp7m R`n{৵3jRfx]Ft<^>6dCC,y? GZOFɛhS|m0hbybL^aV̶swK fd`D6Vts{J,xk|
Z5#WtrRYf?:hK۩%yJ!. (5/A/ ɉ,|IV) 8j+{bH:Gl Mi*DDWkaKvp»;O XͫTztT;6ZSuM*EJ
U" kQ{V7ter&,%h
Jtv /K~4/i3y݆M
{y/(&徱/#3óXu]:Ug [#4m% p~ 6W&+b`e8Th]*%5R['MN =ho.c :ld{i-چTh
Rx.8pB8os-N%L5(sk+@\dE]QXܯ,| w³)අT x},}qjȸ,n G/g
ġ9+U`sțgIfxmu3Js< Kbx+ LzV?m"~/M^Uf0d g*P,}pσ9Xy{B6e Є33`KG-|)?SS
I8sm(j jaP S x9.V!p4똲g k};b r@J`״ۼ;jD5O*8%ݚ
O*&{)i
A ,Qzl٠txDKpeg_ʃ\m!ij#&Cѐe. r/q }s#AK[y4/\7g+1O4-F] \)h*Mj֢^y\atT2=>8={]mM?ٵqiٮICB|_T艳Лpss/Ie 1DZ D>Y5u NƊ{q Bط 2} gm3H;CCnLC>Q~J CLD%z+{E2ߏ_`Q ߺW=FReVR_
Jo50k{yG 4$7.-¨~`* `" ^Lc3b^뻝b'\u'B; P;`?
!$Pca=X19N3!!s w$x&0Xb۹ɞ(ZF_0CFpC*Vw}34
K;(իz x<L
5*F ZAĄMr.)ý 8g˝ɴ2/=QB
y 4!>i0V1^CM:q| 2tǻiۉ+ ;R5}`v"&A
;uz!z)SpFuݭE{E,6+r_b̈́ F~m A\IhIZ?Ԋ5+ M( Ĵ
#o+F Ϛ75bk5Z(%oLCq oG>
A-X#ټ{uG/~#\;CT;X^ krڄ471_&.%_wo^W XZÉ[e$6>>$vcvdbqSN;~T Uݘ2mrL7j &N
e9
ɧP4-G}U7
qWyH#/(J>]X^<, SPR!@9E ]IY)P;Vv-4=b"Zl !2|)w)L2\Yo9OOwG sգRC,Avp ˎ_#,߅W/R@Cn⻋7McBm ۗ=>/*XSzxȨ*
Z a I8Ŋtas.ԋf(@ƴMy_"6HsT$ׂP
Nj
1u 攧Fݣ4?2ޢL\:0acpӳ4魈L_~) 捡cB/YUlly.w8U57.NJݍbj-^HA v_&xg7sƍ6
a d
ϱU( [[8F<`4wWbȎSO&R QJѧ|Kv!i ggDi37 z*{ AiÀarOՕhEN e3I*r^h=jl4Ls)sP>Ԉg ~bݏ%#PGw J &=hÎ0JJw#j
N k=w?a%b)w-sn,Q/v3?]'=L$ >GfɄ5"NS 5؟y[@Ps l7`P&[].Y3i1ۨi ms"bo)oJeqE5_!w92A?'_Jj?@u[V& LT(6 \r8b]jTSΥ4
N{G [< c
Us &}?:Fm~Cpl븿 Vh&QC®=
nq1Bvwp
UDءiP5pz_;[\ݯ=[hc#7J+&)SWygLXu0Б_TGk[\FMɄk(ΛS6@V X=c oǖoٌ\k(^1 ݏ8Ʋڀ)7%=meSvw7 $67jFi9 ]]V*z^{\[P+zw2Q+ґyPl/+ ?=Ӡ7Ƞ0ߛG7x-+ډF 0E4M0MG-UNV X%*)
NP<Ȟv*[؇:\-m7)A|`>W{ 'Vt҆+9Gl& Z =k^&< ^`*|^/4)8P;C9:\}!'ѠWZpj:܆k <=417mVo/ EzZ^\۞ +U?"$'ɑS:Yb~t@HVR9Զ?L-3sez@\MLE7L^+TG()wOp()k? <=ĭNBw5
ϷxPT3$(QޅQ~~buA-Hh *˕%( =X9 | E Φ&ȈP`^yfqkzr*CA4AkI`A~zh~7]?gOdJJVg N!k:h XiK09>:b43@?X0 iv[#L vԛ/ z9^4@ե`Te
"Ɉ6߃ y hX5XwiwT'k֡_xΊuOcI^9Ձ ŁCeOÝ=$67|ͰIB"KmSCjXyC$(" ^X& yjz\`Ӕn.9,'/(7F ;cJ#Pfz< ==<
5,]>'f5@@
x&U jOW% p[a֔FI}Vne~O7XQ-*,pƠ|G,SMQ(b>",oCX3&s(=)鈐'P|aY"Lh,-& gPEjnrI3f{xockZ7fAOp-7R2z˴ >n'1㌖cdB@}ݝ!5YXZ8<;ݕ7*ղ>fo5 7.C" m4\l%zѮxVpdQk }7/*Z~ KAԬI-2Fu~]4o\hL?}
Km.Z& l
r,V9l>0'>*TR9ۗu~W{_ia/lX ;ԣVvo A#A_O1"aje0
mr+b|XoERW\,c"2 l@ +[U@o1G Ed>벮_MvV*G;7ؒҜZB0TI9c+VӨG@V >ם\މ
x2kEފK`gcL6t
8R?FC"s4i N!D Xg/ ]2#l.MׅaǫxFLX =ok~ AK >F >'sǴB -o. rG;sisɸ>[C')Rq@</-, 'oYyq_Ir,'P
R I]Cҳ |qFqZ"Iэ 6#R&o6,jV/KW%Va7CMMxgQ͂ 6=ȃQ唦/-7(_53
Rh N^$F)W'9kxHZ5Hkno#咕^4TTT|J8&h|yگ®o 4w 6ISq 9m}<46^W놳bhڛJ{۠|5xHjQ dlEicH
P2i80bP/Ny~pe
HA&U)ع[hRo)wU֣V1) Fz SIpLI#8=\e^7UP-&ۀ6|JfGf/-u
;ӑ@q{(:ބq[Mi@my%+)۴B< M8b&*}}Wȩ}g''3z:8i Vh*Ӯ]X뒻@"4@/
f#BHHN`~Tv*3bh-QyC,Ga8%F" mFSrSzeE4 )Ph
z7!B+4 oEf[^T }"ן83+a@A39:c! eFdq 4KR5AuKUJK~1#PTǦӔZaJ[-=rph)DO>p`d!
I'NG _4\R-Tپ .J
JV)&Na21 [x_
ƯRlM Y dv2S3j9_sgȘ3ar CR amIа tΚAD }a{vg8J(TZfy=y O]\##:]);Ŧ@^IezQD*'3p#[4 {߳d6ff`3=CD?\l1Il z&]{y=, =f nTLå(0<{E`f#$HToLKZq, ,=5=hлvGhW3+[M
92
H} DȱP?s[Wsf1Zԗz dOkϳXxn BWwWpڵ
KϺ
CFt-2] ,~Ks. 9j^& _vrLm?&_ؒoaKVyn Lx8ᑇ
%ieLKy:dM؆$ KqM(7UN=? aI-
5O+{ #_7̀ \0ɀIð4n}F4k. =!} |]apf,:N5u%ݝg|9 &
st8˟ <:0BOahi- }ޑ)^wLsR+bC#M O,Doߣ}<JP'
.Hj1%5SV%u,ۯ.nuԝ+_ "/jMsbndǙ6(\ }{+şPFJҫ<Ɓ+A@i0hSEh6_=Q^QzJ3/) CcYD ®Q\/Vչ-ּTP`D݈f:!xl6d'Nx_['mJʪR:-Ԑ? F_j^) {S!d HKpfc A|@ <+
inǽѧ~!:ӊ[L,Z 9~I4A٪q,,#ж^PP,>3ψhvh 479 fxΔM=?- ̥)!un}e,ECBچln
)>fPʇgW3{^}|@פ:G.q.?ϳ >.}?'#ڕ7#%EǞv{˺tM0S3@&[cץ.R7E3RO&P&E~춂ԌO=[JCvsNbK
ZG!@*@ $c&Cm9Tjx*\Fƪwf= 1z}5"#GU2k9{*p. x*+,CL
4'mW]M,ٞiC^K6~CP)
ȓN<-`7A 9 oA [ݽ+փsu* ݒ}>Ap ͍vK֭30U mjЍS7kCXC(9= Y} EaWT$\].x] 1aAFWshc#UhN|벟&Qa]ќϻG0Օ7ex1`ȯho69æec$ٓ _EDUx8 bW௳Tc(!;+"Bk|&+(]9bഄ# XE;IiEoϰٕ`.M_RΌeT73 0Bx[F]< Jt2M$ kp68փ
): cmJl LDR
G [p8ߊ; v@DSmC`ڋES'%f"<'˼Bw2u,k-8͋6^)' iۛ q)^Usrꂡ\`]fI2-fB
@dGT,,z | s}MF)CQC>S/>Ïy_`2ϱE0 S:$Cg2Y ˇS ߗ)`A䕙C4@fҟvV q RhUpԼRRl8qF3/ Vҩ`7AfO\Zs C+\Q
eYz۱$e|-@lU/Ȝ(!A / yiG
n*E$:- L@0ߤX{w"Q%\3~c[ %0
RcE!l1=V2ӧoW&Y8|ͱZF1Oj BTs c
g*R[c@TFI0 ! [.$Mv}Ep $a "r-(,^e23Xf ̫XqM9*+ 2F
P 2WΠ#2Y)/c|
cp.-Cǖ&$s Skg$hokg:4<a\Z;0v PE~JGu
*p[zku0;ʯe TWk<ԡO۹lq/ [hf AwŃ%pt Ɋ1Lj כ ^4#iQf
~[d%P^:HgvM]t@@eKM|^آM9ʇ윊?} onic/ׇ/vxk =NZgrQcMtnGqpE5SƄLq&.x(gVҶ
3qǣ4\[{0 $ `#u0B0/ 0%a43/j!h;qB^ JHfHsS7ki;{2ʰEkHpS=ӟݙ[%/b i /lCIg" & \ZznP^?/W5 !j˴ Xn![A;= %fjr/|Y04WZT<<0X
SxN[u:i2" fc/` V~(̳TƸ,<ê3|*
^.Un8cy1-9y4cȀ;[qꎍ8ؠ"Q>Q?*:pERQNԠrj:LW'< %eMh j}7*62 7漇i Bl3OVx "m [
Ýlq5 J0%?,& .vEF 8f+^ =$%~Ihf$nvQ'$Sfvۇz5Gn\h1m'Nց."'͐4X[#O8J;2bU4lѲs6;,p-OHf'S}HCUG ً48(1Ggj)EF43Or'b9 y!0ohY8hXvQ-?#Y*IFRU|kIS9 ɓ!1PNzd?)nI;dxDQg|k\ O$hLbᩫqQy]NMk
"8q25wG4Ӯk_nl QCͫM.HFVZ+wn-=P
WD4^Ɛ&9K捃tH P;\eIۑRl) 2I>!(`3ur,}` hz=@9xDܒK&ɫA&.5ϛT,
=Q#^9R.<
4jXHRYF %],/V:EK&_MH-%5^:sD:C7 pK5](-][}D N\ -a6zH,5U
!] W< W-&w]* ۂF_]]_Q k/uHdQ&}
@쳕IŦ38
hqkGg d&elODv\ɨ* =~s ڶ%W0`W35|;22 xcN5IئElhէwf. s%b&v` U\r{e< :š?Hh.2:f#exa j7EX0%^rG||WNUuGCA f&3+ӆfNؕ p[u}Qr8$n S3I2) D0T^l?$ lIcIr>@Б xUleesUvZv?i4 ۀWa*jx#"V1
DvQ ߽VIӗzp^5~ >svADo *oU`#? ʜ! @L$]WTSfkY
7M{%Aӧ{&WM{}uLLjKKOaA*e'؛'0|<ܰm).5&_hцj}@uW75=OI2Z6LUciK z礷Bxkg scI:-Fm=بEʪǡgU Ǜɸ:|Nh3|zB+*YDʫ]`HFAI
?];&Yz[BfY3Ρ^]#X8r%C".yIՒ#HcLPtZ.&
cz!+ hhIZcjj栴|קn ? @)}EJth_a.1 #Kp#k0cԅz U@IP B<a1.r gO8#-!ufDIN">_r?$b.>LGVQہoN (r_ZO2˽U`?N, =ڃLpoPQM2?wR:kEHl؉%LK9) RլZ-6zhCz
\% +պ:&u
ҟqݷإo_41z~ +hr)-0^W{&Af-O/>
Ki
qBp.!'ΎԄ $!a@"=FEG/hC t M,*?+}A~hI
HrU 3u;yJXXԧbJaws/R@m}QruQ[tH Y!oT93561} 9^Q>'Ury(㎀(I z\xv/Pq-6с
Xk Y95 nsg zߨˋ0 ʱe %b 3,g
LG0:]MZQ=Y&::eyO+ng7,!ϐ<郋p
pEkNeN p8&j/sp
=Â^@d;qZXx].#`Ȝ
Xi߬r%m=f zl[H[6Eg e <Z7`dCj7@49 ݬQ`[ymuL h՚BC
J։6?5ƇX'
="c/Q^ p0r
%c>.f#]RV,"
ycnE \E2xz ǞRA35c#NUl s~7/%c)b9G0bчcG3g 9cܳd濔A!ɞ͍x٤ޭX n>Nmr#2xbڜ拂yi9κNzUۆfh_q@wD*S2kaN?"[j nD c8P BG ^Z > I Sp@s퀩P\ZmܕYx.wS~}e4
'F|EFD@`x57mZK\
Ra k#W\)4VXLHS iOW otKϠ(8;L:m%q+'oT_][4r0IރD~V,uCDEwJK\,LN cl$[ _o)Vx2Upmv w<nJ_ZNnmE =Nܬt5a˰6>V쫐(wdhds"|ݧ}l )spP
d yYcV qֿv}SƲ®`{Y\4? IcɓP 3tgE o@]S73͝綐bP!%.=Xv`0n
dنF3uNYwCݸ?Sq$G+:GsHǵyXS&1s
OҦ7+/Qv6LU}Խz_
4e Fp9TÂyNn-Oɞ*b) P6 I[4f ̠ T=.*8nk"`G|d,{\ѐG>RY8\D$5 |&ҋٽWY R"&'mU8ߤ~# WYQ ^ loe';vr/Dc ӿ6_1/%1Iݕ5{+Ӆ1^hKQaۺЙQ
C"xТM9x Рআk0Ex!#&ud
gmʆZX,[t K ;9AvNeݹM
(>4CewJd)&`yVS{i)s#CŜ(μJuЇx\GLJnmW{+Pz
P~@pD=h:3dc/ֻ.Sпr&fzL|{, ļ6طIo;uS_Ǝ=.
a H HAQ
Yv[ }ؘPxepb_|7?z+ٞC-92H /I)Ƶ;/W
V #6-[nLNyJڱɦ[Si' r苈܅zI@˞
oWd-SPTs1'J)qg*/r2䎭i sLkBF@M8J EτT-愩byJ
7 P{4zp/ ږ$:C'iS3 u An 1
Kk=?WEN76X+*N $S'G ~'tR2W+RؓdRdQ V
$#.8ĕªxY$ـ?A 9WgߐkxFqmH\|3aI]Ex"* [TiS2]" UӅ5bmbϿCjo0U}BA \ǍP,;&-7xkte#% .dntg: pLi2 (ZȱEm~1Hwiv@2ҦLfȣDRMILw:o/_Ӣ!Ts4z ,>/xCd +kddwze[ ;&>` Dp6 בc4Ҏll'l ٓÁ99ՃQfM d$ryXIGKY7S. 9N z撔OGbY lѫӻg Vʂ.]G$OJa]߿#mg\ρSqœCtX/B,Ѡ_r-4\iQ"QЯlȠr}CW|l-[M_ϋqSpoć[?$sP0oٮ=r58%Iʗ2paNR^6Dk!5 -c s ]LfMХ|p|v9YgT( <.ZwtL-mefbQg\ijd4-] 9RKDbViL5 tx~?LMc XϊgX#
_#<_0DԣӴm ~//j?\ 8cp/'֟n`8i]B/Y0 D#ǝ-@MAԣ)N1&>p^dNě>Fy9?HEqZD !ǚ rºHT0!GU϶ Q7pNڟI~Džk;Sr:j94PhߩM}wJV1xbn97Qs M _iIH-2`0Al
H& \0|=ckmnD25"GNQ~BU[}C@@ ංT>x,w{6MEK*st¹HRd
4 S P@ڱB!lPA=!RS(t!3I9ˉx\O<2Cćx*A@0'@ ;"SZƣcGl`N =O@9(qT n4Xz˃2 aMm1ӇV+rΣN;@#ޟEQDtoԘ#j_ _sB}\_nt S
TzHNw*5tlZR`D8-Hܝ
I>8/I& ɰ~9m)GNBu!jYi"N1R팫#|#Ã
ea:ܚ}v֢'w ䷳
ܬK2$MTPk 8\M1Ojr? s lGwks#l[ hR).4c+N* 2D`Y m*E5oޠt%
u S.ixi5 Z/'
2=>>%%&y:{ǿ l!7sZqSx
n #ףpw|eFK]IЮ#ZI3]yDYKo9Y̬wA-˞HlYU3 >%JT Pj⽳$˕:'23.R wDF@]B#Lupk!K| xM!MH]-((JD+Jhx;u1 v?&UHʜhL
zsl%EI[MF,ǃ@L*iv,*(|M;oyz7[T?:dڤH I%}xV`\v b/Ma@~Y?saiۃ ˵^Eg WxlX
Z3L @{jQ{-,r-'cFi\EȞb$]xNj6X8H[KK8y7V1!# NѰ©hdGI?ܖDS3K
Zev?_ tY.d:9%ڗj⃕Q#R ~gfkZYHqSR*t\2qC۽ +$L *.h؎?T\ a)>Z3!i F&HI L̢_IR4iuq7@\/$8!>J?n
&vsN1
³Ex"hfn[$XecT 8jC.<]m@iNX-,TJf(GraQCkҷIe7X^AEiC< >'m J8ߥ" rg! ƯG;ڝ=D.2.azΕJ@6"͓;
4dy sU^nV"vî e
^z&4
<ٶ _)Kwש %h&HZ' W!܀rej#Ы9OJ92
a(H.*zoZ vM9/lHo !-=#ܼٝM'
ww"ixZiF|sv܇
ɹGbe9Yd 7 cyAtsAe#?tV/$e?^yr1 I6=kY*ײkYTsSKRD? ^Ɛw뷜F8&W!Fw3ś
p# UR y] F+m#T glvJ;[$AR"ƃ V)BI qMY)hzTULgrEJc- f _ŔqR>OYiFDnA۲<ۑ Qe9'>"Ă6zmWٖ :Ii"q R)yW8y(&qpaF֬ˌeѡ߃C eqIڻrHT) qnyxgoN0ǐSsk ul,l'\a DH.6x\516j9&FЈKҬ)ޗA%Uu2m2JCѨI0.U ^m ?vH+93k}v_dŰ8c>ܕ =
:|iR3K}V% Rjs!ctN#!I,RB/: Z 2,{u Sj@, S8gYJ %vx;_9Fyv}K͑e4вe.XRd&wj 9"@l)okgG^7 ?iuxF'Zl ):~cdV'Ӡ'o9[@~VZuSI,eBodĻ( iϖ
su1~-;jO7
e~9<RLsL \5"֧Z] 3ݛQ2N^*i xp!ŌT$BbN*Q U1q=8P9ռZ3SmY7:|z+_1FaMK2ϕ%܅{E #D:bT٧i 8YZ1{JrY^AcR5i'!rviw y?wE ØFGU3'8gDE4#[VUG >o\?h8Op;'u o871G dKd/7\>'+wkH *QntϬ4}d'bSZY+F 3P"<U`cg)]Iqe:? i\(Aoj,jDv&f +.A֛ Oj&o:c!
/g&3Y jaOn6;QE4OT_7EH3 (RZ^~pg2P.܀j(z}oT%U8 o(X^. L\%5
fLME $ " :# z<"
RL/#қLq3+ !ᅊڤs][j=nL }҄Ǜ0kmP 6 ^姎B1̂=mP7aVGg*7~`4tP
4L%J JP7
El5d ~8Li08.jDSwJōaPn%q8n?-Yȟ*AW-נ(cCdu+D֪ X4S?/ȕ̯m*Z$|w/v6H[f{X5G˹DBMb1;wh$$y1~Nbg H(H ~xJ!8V [ @uhXn >7o!YKXd͡
h/<<[!b&vѴ:ӿdB,/=mBrp(cJz`(\wl룗L)5^AۻZIBYt6aQorN-~ۑXZW?X( Ysfun}{o'\GVS%(OJi}-
~"ZIm
':q|&[}id>HF'%$sWN?g\vD鵎ʹ">2/Ȩ=\SޙMך ˠuY9q&?q@O c$^c瘹OwێDuG0Tq\ LɠOѣڢU079՚):Gv /\ݒ Ce /joaB3eM/^wu1_>2 + wb\Ij X: $~]Hh!~7 e'S"DkKI>[L 8M A@,d oí+7Vz 8MkѰl
#Qާ?3But
C㐲9ƍ
)#ߴE5YXd*-Z"pƉ낎
jpycPA#p
EsdW |ZN& ШӢ9 j#,v7کgH ·
cPH.Ă+%:Z:býFP6d9Ǒ\V#eۅ7`hz #d ߚJ e܊,Av<7r1VpwDj?a-L#N1[u&i'.p`%<.g#p m/xOQjݻ/fsqeo.# 5+&MA3l 552AogM*pAFDzvONN@a]n `)sEZ)=s4)v1B]E{ЃgrkjCpB?-NR/:L/~_|8"!d>ծ,)W'/Ԏh
51w3qNt!db׃Z!"xN m6Zq=EFOKpК^'%BRILqZ,.
Y=bB~9Q[WEڡUX.NlR4\=Aped1 -3<@g ůGY+FNSc ΄F xJzn A6 h+Iqؼ!QC )=|`+O%րamWg:*N ,26m _7w8Y 7ӋUfPUF~\aM{-1=-!PWeuzW7v0`<8ݮK1ఄI$+GrŲR ڇŁU0 Oa'}q&ȷB81xa Lb0,mes\CJ8|΅Hvӡ.)/AʽaZ,/L1jV 5p9}udu%L[ZyXto ҉5 )D%Y߃Ol=N\:!=ג hϑhέ D&4l&Ep <οiFB ]l ~"`ƌ/8o^^z3 ,GD<¨ [8>ǵdt((MR+{7sYo8g%ۆE)-Y,c:xjb8M T9 b5uGMWB@c]ZC7;B7
`EGK)v=-"">"Ć7!s F]#$7 ̽Z*G&lK#ګWsܾZŕ?oӗQo
R#2 隱" 朇hB// VO?Ȓ )108(rCs) Ȋ9GiGk
X^n7n^EϱԹrr-h&%˶3+5k}vwmJ0dX&. rdR%i$:gEx(G1eM2o"Ɓ
j\ +aJV Q=v@ KgZ5Y>@k[=.YA<ݼVpO%a?wOn_Q
"hƉ~Sx-!_c`y
yJ2}ɕPY`ѥ}&5Zʇ$b Kֻl袎6y0wUt6@tf;}JģwŊAB`Ý "ހZ7B*V4IQIOS<ÕSz%u@wħ3 fe/i3
-dop ji`|-5&!sڳǽ#2KeENCBA 5d!0)j>8?ٟ~Qۻ] qt$j
r\hOc=mQEh Z?kcw{U2Z/ʍeaWsK0>*KUex0P>NՆƵ ۴VoB9R^ТĢ!hxyרobm mi45ͧSSO@ߑ=6iyk)ID 8$*'JFAcx.}|"1$,-_ 'ȷ&H*ϊULkdsOxCy|SFnK`3
;B&t72
8eUk>GɫLp"dJbѝd.4@fJ6=5ZZ d^%˙֣ Bf(hyއ[;fp;չV|ЪFw\U{4Y(]]le3P ɸ-01B Bt_M*S k%_6h#t|R>U
| ⶌ_/pEɩ,mZ>'kQ ,^FU]\
^'J¦ӁXX $7 !I'G
b$,qU&IKl Z2
UK5|Z w![\=yܜF{ Y%WB,;eiY5%[6 6iRĭ #(>YĢ&#P54s/)rOսG rxx=̌n6e E݃Ӯh
exnl4%#\ZXrݍ+[.+ju9`0OE rcmCҫI [-ۉy,:p7ĿK #%oZDluo 4BS2"ۘ ^n̰?38Cs̖5k;Rr/ĄbŶEE HHKr~sA3?%ˉRjG@E g7>b!CUΠ՞77X %%ލeMC|i%Y |\7ʘӰ{R1w E䃵G-P̄ADj=4;Z>Mu26(f3M\wcv5?Q%.[=T۶/1O Ms-I Q. 9EG:uՠʀ7&P*㔴eqE"|\3aQ3/ƣq⒕NuT\7Q(-h){`6oX& WI+clS/ K MS{u
{ F {tv@ dl.X'Bߡ*7Mw,ZZwQh!O+Hm1-2GX 2pcM?p*M5" %롖4tS vtݳ+"&{-T[hD@,aEJ^7sI :Wz0x
FL h~l"u|'x'uǹUVԌ|4F%ЁJ7F| c
Y9(4uW\]=r)f)DO!=K X @/:RaqUNzGeFUQxw8Vf w>[n*DN|Wuud싗!ʇ'_ |6fDP,e f$U J#$T]T)ժf\5DA
)6nYِȵ.#HXE;5RM̹]Uisvmrg\UK$+j
m7=X$L-k[!O"S`1eDF+Fu!wf?> ~:Xbq +"}9</Ǯmߊ- qF1K2=z=IcH7.F,UԯWnFCըԻ&=z6c\3|)ƭmw'KxXdɽFhǤNk Gw:Ygmg҂U7 - zF9]B )EME F98e巟0j * 7{p1 ڙ,e&&n
z/} MHٝ d/UR2Ln0YPvkS|h!~tSAHS4&We GߒI7Q SK(v``& <4v~ Hzh>) EgLDC:
|[ +
f;u| oWN(_q
4)fx%uM5`4fAXzCq^Q3=Wi{ U`$# ߭Y _h_ml@`=̛ѴH8~HO"05 l scԘ
ؔ24$mȘ-6 ٹJΈ}~ȤeiEŅA=1MNm b@f ҶLE D#ZerrHE|],܁
YZ