Recherche avancée

Sur d’autres sites (4084)

• dnn : add openvino as one of dnn backend

  25 mai 2020, par Guo, Yejun

  dnn : add openvino as one of dnn backend
  OpenVINO is a Deep Learning Deployment Toolkit at
  
  https://github.com/openvinotoolkit/openvino, it supports CPU, GPU
  
  and heterogeneous plugins to accelerate deep learning inferencing.
  Please refer to https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md
  
  to build openvino (c library is built at the same time). Please add
  
  option -DENABLE_MKL_DNN=ON for cmake to enable CPU path. The header
  
  files and libraries are installed to /usr/local/deployment_tools/inference_engine/
  
  with default options on my system.
  To build FFmpeg with openvion, take my system as an example, run with :
  
  $ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH :/usr/local/deployment_tools/inference_engine/lib/intel64/ :/usr/local/deployment_tools/inference_engine/external/tbb/lib/
  
  $ ../ffmpeg/configure —enable-libopenvino —extra-cflags=-I/usr/local/deployment_tools/inference_engine/include/ —extra-ldflags=-L/usr/local/deployment_tools/inference_engine/lib/intel64
  
  $ make
  Here are the features provided by OpenVINO inference engine :
  
   support more DNN model formats
  
  It supports TensorFlow, Caffe, ONNX, MXNet and Kaldi by converting them
  
  into OpenVINO format with a python script. And torth model
  
  can be first converted into ONNX and then to OpenVINO format.
  see the script at https://github.com/openvinotoolkit/openvino/tree/master/model-optimizer/mo.py
  
  which also does some optimization at model level.
   optimize at inference stage
  
  It optimizes for X86 CPUs with SSE, AVX etc.
  It also optimizes based on OpenCL for Intel GPUs.
  
  (only Intel GPU supported becuase Intel OpenCL extension is used for optimization)
  Signed-off-by : Guo, Yejun <yejun.guo@intel.com>
  
  Signed-off-by : Pedro Arthur <bygrandao@gmail.com>
  

  • [DH] configure
  • [DH] libavfilter/dnn/Makefile
  • [DH] libavfilter/dnn/dnn_backend_openvino.c
  • [DH] libavfilter/dnn/dnn_backend_openvino.h
  • [DH] libavfilter/dnn/dnn_interface.c
  • [DH] libavfilter/dnn_interface.h

• Error:No such property : targetPlatform for class : com.android.build.gradle.managed.NdkConfig

  30 juillet 2016, par Alder

  I am trying to build FFMPEG into my JNI code with gradle in Android Studio. I have build FFMPEG as a .so file, in order to adapt different platform, I build it for different ABI(arm64-v8a, armeabi-v7a, mip, etc).Then I need to determine the ABI of the current build in the build.gradle file.

  Refer Experimental Plugin User Guide, my build.gradle look like this :

  apply plugin: 'com.android.model.native'    
  
  model{
  
      repositories {
  
          prebuilt(PrebuiltLibraries){
  
              ffmpeg{
  
                  headers.srcDir "src/main/jni/build/${targetPlatform.getName()}/include"
  
                  binaries.withType(SharedLibraryBinary) {
  
                      sharedLibraryFile = file("src/main/jni/build/${targetPlatform.getName()}/libvflibrary.so")
  
                  }
  
              }
  
          }
  
      }
  
      android {
  
          compileSdkVersion = 24
  
          buildToolsVersion = "23.0.3"
  
  
  
          defaultConfig {
  
              minSdkVersion.apiLevel = 15
  
              targetSdkVersion.apiLevel = 24
  
              versionCode = 1
  
              versionName = "1.0"
  
          }
  
  
  
          ndk{
  
              //platformVersion = 21
  
              moduleName = "library-jni"
  
              stl = 'gnustl_static'
  
              toolchain = "clang"
  
              abiFilters.addAll(['armeabi', 'armeabi-v7a', 'arm64-v8a', 'mips'])
  
              cppFlags.addAll(['-std=c++11', '-D__STDC_CONSTANT_MACROS'])
  
              ldLibs.addAll(['log', 'android', 'z', 'EGL', 'GLESv2'])
  
          }
  
  
  
          sources {
  
              main {
  
                  jni {
  
                      source{
  
                          srcDirs 'src/main/jni'
  
                      }
  
                      dependencies {
  
                          library 'ffmpeg' linkage 'shared'
  
                      }
  
                  }
  
              }
  
          }
  
      }
  
  }

  I am getting an error :

  Error:No such property : targetPlatform for class :
  com.android.build.gradle.managed.NdkConfig.

  Does anyone have an idea on how I can solve this, please ?

• CMTimeGetSeconds doesn't get the right video duration

  26 février 2015, par JLCastillo

  some users don’t get the right duration of videos they capture with their own device. The funny thing is others do actually see it right, using the same device models and OS version. Anyway, we observed it in a iPhone 5c 7.1.2 and an iPhone 5s 8.1.3.

  This code works for most users, but not all :

  ALAssetRepresentation *representation = [mediaObject.asset defaultRepresentation];
  
  NSURL *url = [representation url];
  
  NSDictionary *options = @{ AVURLAssetPreferPreciseDurationAndTimingKey : @YES };
  
  AVAsset *avAsset = [AVURLAsset URLAssetWithURL:url options:options];
  
  videoDurationTime = CMTimeGetSeconds(avAsset.duration);

  I asked them to send the input videos, and this is the output from "ffmpeg -i"

  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'bug_duration1.MOV':
  
    Metadata:
  
      major_brand     : qt
  
      minor_version   : 0
  
      compatible_brands: qt
  
      creation_time   : 2015-02-23 08:30:01
  
      encoder         : 8.1.3
  
      encoder-eng     : 8.1.3
  
      date            : 2015-02-23T16:30:01+0800
  
      date-eng        : 2015-02-23T16:30:01+0800
  
      model           : iPhone 5s
  
      model-eng       : iPhone 5s
  
      make            : Apple
  
      make-eng        : Apple
  
    Duration: 00:00:03.67, start: 0.000000, bitrate: 16793 kb/s
  
      Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 1920x1080, 16719 kb/s, 29.99 fps, 29.97 tbr, 600 tbn, 1200 tbc (default)
  
      Metadata:
  
        creation_time   : 2015-02-23 08:30:01
  
        handler_name    : Core Media Data Handler
  
        encoder         : H.264
  
      Stream #0:1(und): Audio: aac (mp4a / 0x6134706D), 44100 Hz, mono, fltp, 61 kb/s (default)
  
      Metadata:
  
        creation_time   : 2015-02-23 08:30:01
  
        handler_name    : Core Media Data Handler

  The video is detected with a duration of several minutes. Did anybody face this problem before ?

  Thanks in advance.