标题: rknn 1.6 模型转换失败 [打印本页] 作者: marshorizon 时间: 2021-3-6 10:00 标题: rknn 1.6 模型转换失败 环境：rknn 1.6 docker x86，
自己在docker中训练个模型，转换rknn模型时失败。信息如下，劳驾指点一下。
D Save log info to: rknn.log
I Start importing keras...
D Convert Keras with Keras Branch
D import clients finished
I build output layer attach_fc1000/Softmaxut0
I Try match fc1000
I Try match avg_pool
I Try match activation_48
I Try match add_15
。。。
。。。
。。。
D Merge ['conv1_176', 'bn_conv1_175'] (convolution)
D Merge ['conv1_pad_177', 'conv1_176'] (convolution)
D Optimizing network with conv2d_big_kernel_size_transform
D Optimizing network with auto_fill_zero_bias, auto_fill_tf_quantize, align_quantize, broadcast_quantize, qnt_adjust_coef, qnt_adjust_param
I End importing keras...
E RKNN model data is None, please load model first!
E RKNN model data is None, please load model first!
ut0