使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败

hhdy · 发表于 2023-4-20 11:45:05

本帖最后由 hhdy 于 2023-4-20 11:45 编辑

转换脚本（Demo中的 onnx 模型可以成功转换）：

import os
import urllib
import traceback
import time
import sys
import numpy as np
from rknn.api import RKNN
ONNX_MODEL = 'test4.onnx'
RKNN_MODEL = 'test.rknn'
if __name__ == '__main__':
rknn = RKNN(verbose=True)
# pre-process config
print('--> Config model')
rknn.config(target_platform='rk3588', optimization_level=0)
print('Config model done')
# Load ONNX model
print('--> Loading model')
ret = rknn.load_onnx( model=ONNX_MODEL ) #, inputs=['inputs', 'buffer'], input_size_list=[[514],[27724] ], outputs=['out', 'out_buffer'])
if ret != 0:
print('Load model failed!')
exit(ret)
print('Loading model done')
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=False)
if ret != 0:
print('Build model failed!')
exit(ret)
print('Building model done')
# Export RKNN model
print('--> Export rknn model')
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export rknn model failed!')
exit(ret)
print('Export rknn model done')
rknn.release()

复制代码

错误信息：

W __init__: rknn-toolkit2 version: 1.4.0-22dcfef4
--> Config model
Config model done
--> Loading model
W load_onnx: The config.mean_values is None, zeros will be set for input 0!
W load_onnx: The config.std_values is None, ones will be set for input 0!
W load_onnx: The config.mean_values is None, zeros will be set for input 1!
W load_onnx: The config.std_values is None, ones will be set for input 1!
Loading model done
--> Building model
E build: Catch exception when building RKNN model!
E build: Traceback (most recent call last):
E build: File "rknn/api/ir_graph.py", line 78, in rknn.api.ir_graph.IRGraph.infer_shapes
E build: File "/home/hhdy/.local/lib/python3.6/site-packages/onnx/shape_inference.py", line 37, in infer_shapes
E build: inferred_model_str = C.infer_shapes(model_str, check_type, strict_mode)
E build: onnx.onnx_cpp2py_export.shape_inference.InferenceError: [ShapeInferenceError] (op_type:Reshape, node name: _0_expand1): [ShapeInferenceError] Inferred shape and existing shape differ in dimension 2: (1) vs (5)
E build: During handling of the above exception, another exception occurred:
E build: Traceback (most recent call last):
E build: File "rknn/api/rknn_base.py", line 1546, in rknn.api.rknn_base.RKNNBase.build
E build: File "rknn/api/graph_optimizer.py", line 1335, in rknn.api.graph_optimizer.GraphOptimizer.fuse_ops
E build: File "rknn/api/ir_graph.py", line 343, in rknn.api.ir_graph.IRGraph.rebuild
E build: File "rknn/api/ir_graph.py", line 185, in rknn.api.ir_graph.IRGraph._clean_model
E build: File "rknn/api/ir_graph.py", line 76, in rknn.api.ir_graph.IRGraph.infer_shapes
E build: File "rknn/api/ir_graph.py", line 84, in rknn.api.ir_graph.IRGraph.infer_shapes
E build: File "/home/hhdy/.local/lib/python3.6/site-packages/onnx/shape_inference.py", line 37, in infer_shapes
E build: inferred_model_str = C.infer_shapes(model_str, check_type, strict_mode)
E build: onnx.onnx_cpp2py_export.shape_inference.InferenceError: [ShapeInferenceError] (op_type:Reshape, node name: _0_expand1): [ShapeInferenceError] Inferred shape and existing shape differ in dimension 2: (1) vs (5)
Build model failed!

复制代码

hhdy · 发表于 2023-4-20 11:46:27

如果解决问题，需要我的模型，也可以提供

板蓝根 · 发表于 2023-4-20 16:11:18

提供一下模型

hhdy · 发表于 2023-4-21 09:52:23

模型

hhdy · 发表于 2023-4-21 11:15:49

板蓝根发表于 2023-4-20 16:11
提供一下模型

大神模型发出来了。你看下

hhdy · 发表于 2023-4-21 19:15:19

本帖最后由 hhdy 于 2023-4-21 20:01 编辑

+++++

icy · 发表于 2023-7-24 15:41:50

问题解决了么？

[技术讨论] 使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败

【技术讨论】使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败

优秀版主

[技术讨论] 使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败

【技术讨论】 使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败

优秀版主

【技术讨论】使用 rknn-toolkit2，我的 onnx 模型转换 rknn 模型失败