Tensorflow对象检测API索引超出界限

Question

运行Android TF Detect Demo并使用我重新训练的3类ssd_mobilenet_v1_coco模型时，我的TF检测Demo崩溃，给出 IndexOutOfBoundException ，这是

  

12-26 17：53：13.931 22429-25212 / org.tensorflow.demo E / AndroidRuntime：FATAL EXCEPTION：推理                                                                        处理：org.tensorflow.demo，PID：22429                                                                        java.lang.ArrayIndexOutOfBoundsException：length = 3;索引= 3                                                                            at java.util.Vector.arrayIndexOutOfBoundsException（Vector.java:907）                                                                            在java.util.Vector.elementAt（Vector.java:328）                                                                            在java.util.Vector.get（Vector.java:442）                                                                            at org.tensorflow.demo.TensorFlowObjectDetectionAPIModel.recognizeImage（TensorFlowObjectDetectionAPIModel.java:194）                                                                            at org.tensorflow.demo.DetectorActivity $ 3.run（DetectorActivity.java:289）                                                                            在android.os.Handler.handleCallback（Handler.java:739）                                                                            在android.os.Handler.dispatchMessage（Handler.java:95）                                                                            在android.os.Looper.loop（Looper.java:148）                                                                            在android.os.HandlerThread.run（HandlerThread.java:61）

来自

beans = {
    userDetailsService(Your_service_name)
}

Answer 1

添加“???”解决我的Label文件中的字符串，似乎应用程序遵循第一个ID为1的pbtxt格式。