当我重用模型并尝试使用图像进行预测时,它会抛出一个错误,说明它是不兼容的。print(CATEGORIES[int(prediction[0][0])])WARNING:tensorflow:Model was constructed with shape (4, 112, 112, 3) for input Ker
我正在尝试使用我的Keras模型和我的TF对象检测模型。但是我得到了这个错误: ValueError: Calling `Model.predict` in graph mode is not supported when the `Model` instancePlease construct your `Model` instance in graph mode or call `Model.predict` with eager mode enabled.detection_graph.as_default():
-in-keras/)。在下面的代码块中,我试图“做出预测”: # make prediction# summarize the shape of the listfor a in yhat]) 我收到以下错误: --------------------------------------------------------------------------<ipython-input-35-278
# make predictions下面是对不可见时间步骤的实际预测当这里的batch_size设置为1时,错误是:
ValueError: Cannot feed value of shape(1, 1, 2) for Tensor 'lstm_1_input:0', which has shape '(10, 1