Tool/software:
Hello, I'm trying to import custom model using edgeai-tidl-tools.
My model looks llike below image.
And i got error message like this
root@5fcd978baa80:/home/root/examples/osrt_python/tfl# python3 tflrt_delegate.py -c Running 1 Models - ['dimension_test'] Running_Model : dimension_test delegate setting start tidl_tools_path = /home/root/tidl_tools artifacts_folder = ../../../custom_model/artifacts/dimension_test/ tidl_tensor_bits = 8 debug_level = 5 num_tidl_subgraphs = 16 tidl_denylist = tidl_denylist_layer_name = tidl_denylist_layer_type = tidl_allowlist_layer_name = model_type = tidl_calibration_accuracy_level = 7 tidl_calibration_options:num_frames_calibration = 5 tidl_calibration_options:bias_calibration_iterations = 3 mixed_precision_factor = -1.000000 model_group_id = 0 power_of_2_quantization = 3 ONNX QDQ Enabled = 0 enable_high_resolution_optimization = 0 pre_batchnorm_fold = 0 add_data_convert_ops = 3 output_feature_16bit_names_list = m_params_16bit_names_list = m_single_core_layers_names_list = Inference mode = 0 Number of cores = 1 reserved_compile_constraints_flag = 1601 partial_init_during_compile = 0 ti_internal_reserved_1 = delegate setting fin! Model path : ../../../custom_model/model/new_prelu_to_mul_cut5.tflite compile mode! ========================= [Model Compilation Started] ========================= Model compilation will perform the following stages: 1. Parsing 2. Graph Optimization 3. Quantization & Calibration 4. Memory Planning ============================== [Version Summary] ============================== ------------------------------------------------------------------------------- | TIDL Tools Version | 10_00_04_00 | ------------------------------------------------------------------------------- | C7x Firmware Version | 10_00_02_00 | ------------------------------------------------------------------------------- ============================== [Parsing Started] ============================== [TIDL Import] [PARSER] WARNING: Network not identified as Object Detection network : (1) Ignore if network is not Object Detection network (2) If network is Object Detection network, please specify "model_type":"OD" as part of OSRT compilation options [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- model_5/model_4/model_3/batch_normalization_132/FusedBatchNormV3;model_5/model_4/model_3/conv2d_81/BiasAdd/ReadVariableOp/resource;model_5/model_4/model_3/conv2d_81/BiasAdd;model_5/model_4/model_3/depthwise_conv2d_68/depthwise;model_5/model_4/model_3/conv2d_81/Conv2D -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 5 Tflite layer type --- 18 layer output name--- customed_output_first -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 5 Tflite layer type --- 18 layer output name--- customed_output_third -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 3 Tflite layer type --- 19 layer output name--- customed_output_fourth -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- model_5/model_4/model_3/p_re_lu_126/add;model_5/model_4/model_3/p_re_lu_126/Relu;model_5/model_4/model_3/p_re_lu_126/Neg_1;model_5/model_4/model_3/p_re_lu_126/Relu_1;model_5/model_4/model_3/p_re_lu_126/mul1 -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- model_5/model_4/model_3/batch_normalization_133/FusedBatchNormV3;model_5/model_4/model_3/depthwise_conv2d_63/depthwise;model_5/model_4/model_3/conv2d_82/Conv2D -- [tidl_tfLiteRtImport_core.cpp, 3090] [TIDL Import] [PARSER] SUPPORTED: Supported TIDL layer type --- 5 Tflite layer type --- 18 layer output name--- customed_output_first -- [tidl_tfLiteRtImport_core.cpp, 3090] Total Nodes = 7 ------------------------------------------------------------------------------- | Core | No. of Nodes | Number of Subgraphs | ------------------------------------------------------------------------------- | C7x | 7 | 1 | | CPU | 0 | x | ------------------------------------------------------------------------------- ============================= [Parsing Completed] ============================= In TIDL_tfliteRtImportInit subgraph_id=490 Layer 0, subgraph id 490, name=customed_output_first Layer 1, subgraph id 490, name=input_12 In TIDL_tfliteRtImportNode, TIDL Layer type - 1, Tflite builtin code type - 3 In TIDL_tfliteRtImportNode, TIDL Layer type - 5, Tflite builtin code type - 18 In TIDL_tfliteRtImportNode, TIDL Layer type - 5, Tflite builtin code type - 18 In TIDL_tfliteRtImportNode, TIDL Layer type - 3, Tflite builtin code type - 19 In TIDL_tfliteRtImportNode, TIDL Layer type - 5, Tflite builtin code type - 0 In TIDL_tfliteRtImportNode, TIDL Layer type - 1, Tflite builtin code type - 3 In TIDL_tfliteRtImportNode, TIDL Layer type - 5, Tflite builtin code type - 18 =================== [Optimization for subgraph_490 started] =================== In TIDL_runtimesOptimizeNet: LayerIndex = 9, dataIndex = 8 [TIDL Import] [PARSER] WARNING: Batch Norm Layer customed_output_third's coeff cannot be found(or not match) in coeff file, Random bias will be generated only for evaluation usage. Results are all random [TIDL Import] ERROR: - Failed in function: tidl_optimizeNet -- [tidl_import_core.cpp, 2473] [TIDL Import] ERROR: Network Optimization failed - Failed in function: TIDL_runtimesOptimizeNet -- [tidl_runtimes_import_common.cpp, 1232] interpreter created! batch : 1 new_height : 256 new_width : 256 channel : 3 Segmentation fault (core dumped)
1. How can i solve this error?
2. How can I find the "tidl_import_core.cpp" or "itdl_runtimes_import_common.cpp" or the orignal src file of "tidl_model_import_tflite.so"?
thank you.
