No implementation found for java.lang.String org.tensorflow.lite.TensorFlowLite.version()

编译tensorflow so 步骤：

• 通过运行tensorflow$> sudo ./tensorflow/contrib/lite/java/build_aar_for_release.sh 编译so
• 然后so(路径： bazel-bin/tensorflow/contrib/lite/java/libtensorflowlite_jni.so)拷贝到自己项目中

运行项目报错：

Process: com.android.tcl.tensorflowlite, PID: 3479
    java.lang.UnsatisfiedLinkError: No implementation found for java.lang.String org.tensorflow.lite.TensorFlowLite.version() (tried Java_org_tensorflow_lite_TensorFlowLite_version and Java_org_tensorflow_lite_TensorFlowLite_version__)
        at org.tensorflow.lite.TensorFlowLite.version(Native Method)
        at com.example.android.tflitecamerademo.ImageClassifier.<init>(ImageClassifier.java:174)
        at com.android.tcl.tensorflow.RecognizeTensorFlow.onCreate(RecognizeTensorFlow.java:158)

将手机程序log输出到文件中，查找发现

10-11 13:42:20.092  1341  1341 W System.err: TensorFlowLite: failed to load native library: dlopen failed: "/data/app/com.android.tcl.tensorflowlite-77WtJMHLi09M1fD_M0e_fw==/lib/arm/libtensorflowlite_jni.so" is 64-bit instead of 32-bit
10-11 13:42:20.092  1341  1341 E zygote  : No implementation found for java.lang.String org.tensorflow.lite.TensorFlowLite.version() (tried Java_org_tensorflow_lite_TensorFlowLite_version and Java_org_tensorflow_lite_TensorFlowLite_version__)

原来是由于so版本不对，手机是32位机，so却是64位。原来之前获取so的路径不正确，编译so成功后会在tensorflow根目录
下生成一个tflite-1.0.aar文件，解压文件进入jni下：

x86_64是64位机so
armeabi_v7a是32位so