Toybrick

标题: unet模型转为rknn模型后rknn.init_runtime()出现问题 [打印本页]

作者: zhanglao    时间: 2020-8-5 16:21
标题: unet模型转为rknn模型后rknn.init_runtime()出现问题
onnx的unet模型成功转为rknn模型后,pc端模拟推理时在初始化runtime时出现问题: [setup_nn_instancenorm:2613]shape of nn_param.instancenorm.axis maybe error!
详细错误如下:
--> Loading RKNN model

--> Load RKNN model succeed


--> Init runtime environment

D target set by user is: None
D [vnn_CreateRKNN:5114]current chip platform = x86

W [set_chip_platform_env:187]evaluate model on RK1808
I [vnn_CreateRKNN:5318]rknn model optimization_level=3
I [setup_client_ops:4866]setup_client_ops num=0
I [setup_client_ops:4991]setup_client_ops finish
I [register_client_ops:5010]register_client_ops num=0
I [register_client_ops:5028]register_client_ops finish
I [setup_nodes:4093]name=Conv_28_26
I [setup_nodes:4093]name=LeakyRelu_29_38
I [setup_nodes:4093]name=Conv_30_32
I [setup_nodes:4093]name=LeakyRelu_31_31
I [setup_nodes:4093]name=Conv_32_25
I [setup_nodes:4093]name=Add_33_21
I [setup_nodes:4093]name=LeakyRelu_34_17
I [setup_nodes:4093]name=InstanceNormalization_37_57
E [setup_nn_instancenorm:2613]shape of nn_param.instancenorm.axis maybe error!
I [remove_client_ops:5039]remove_client_ops num=0
I [remove_client_ops:5050]remove_client_ops finish
Create RKNN model fail!
E Catch exception when init runtime!
E Traceback (most recent call last):
E   File "rknn/api/rknn_base.py", line 1067, in rknn.api.rknn_base.RKNNBase.init_runtime
E   File "rknn/api/rknn_runtime.py", line 326, in rknn.api.rknn_runtime.RKNNRuntime.build_graph
E Exception: RKNN init failed. error code: RKNN_ERR_FAIL
--> Init runtime environment failed

弄了几天无果,寻求帮助!

作者: leok    时间: 2020-8-6 09:09
optimization_level=2   试试看。   
NPU里头精度最高的MAC是fp16。




欢迎光临 Toybrick (https://t.rock-chips.com/) Powered by Discuz! X3.3