最近再加一个人脸识别得功能,转换facenet的tf模型到paddle,报错,请指教
Created by: baigang666
我的tf模型来自https://github.com/davidsandberg/facenet.git 此模型再tf==1.7.0上可以正常运行 使用x2paddle转换到paddle模型时 按照提示,输入 Unknown shape for input tensor[tensor name: "batch_size"] Use your keyboard type the shape of input tensor below :) Shape of Input(e.g. None,224,224,3): 1,160,160,3 Unknown shape for input tensor[tensor name: "phase_train"] Use your keyboard type the shape of input tensor below :) Shape of Input(e.g. None,224,224,3): 1,160,160,3
之后报错: ValueError: Shape must be rank 0 but is rank 4 for 'InceptionResnetV1/Conv2d_1a_3x3/BatchNorm/cond/Switch' (op: 'Switch') with input shapes: [1,160,160,3], [1,160,160,3].
请大神帮忙看一下,谢谢