This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: 我使用edge-ai-tools工具编译了一个车道线检测模型,部署在EVM上时,准确率降低了很多,这是什么原因导致的啊?

Part Number: TDA4VM

我使用edge-ai-tools工具编译了一个车道线检测模型,部署在EVM上时,准确率降低了很多,这是什么原因导致的啊?

  • 请问是自己的板子还是TI的EVM板? 具体编译步骤是怎么样的?能否贴一下代码和log输出信息。

  • TI的EVM板。就按照edge-ai-tools工具里的readme编译的。log如下。

    Available execution providers : ['TIDLExecutionProvider', 'TIDLCompilationProvider', 'CPUExecutionProvider']

    Running 1 Models - ['bestb']


    Running_Model : bestb


    Running shape inference on model ../../../models/public/culane_18.onnx

    --------------------2---------------------------------
    tidl_tools_path = /home/leo/code/edgeai-tidl-tools-master/examples/osrt_python/ort/tidl_tools
    artifacts_folder = ../../../model-artifacts//bestb/
    tidl_tensor_bits = 8
    debug_level = 1
    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 = 2
    tidl_calibration_options:bias_calibration_iterations = 5
    mixed_precision_factor = -1.000000
    model_group_id = 0
    power_of_2_quantization = 2
    enable_high_resolution_optimization = 0
    pre_batchnorm_fold = 1
    add_data_convert_ops = 3
    output_feature_16bit_names_list =
    m_params_16bit_names_list =
    reserved_compile_constraints_flag = 1601
    ti_internal_reserved_1 =

    ****** 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******

    Supported TIDL layer type --- Conv -- Conv_0
    Supported TIDL layer type --- Relu -- Relu_1
    Supported TIDL layer type --- MaxPool -- MaxPool_2
    Supported TIDL layer type --- Conv -- Conv_3
    Supported TIDL layer type --- Relu -- Relu_4
    Supported TIDL layer type --- Conv -- Conv_5
    Supported TIDL layer type --- Add -- Add_6
    Supported TIDL layer type --- Relu -- Relu_7
    Supported TIDL layer type --- Conv -- Conv_8
    Supported TIDL layer type --- Relu -- Relu_9
    Supported TIDL layer type --- Conv -- Conv_10
    Supported TIDL layer type --- Add -- Add_11
    Supported TIDL layer type --- Relu -- Relu_12
    Supported TIDL layer type --- Conv -- Conv_13
    Supported TIDL layer type --- Relu -- Relu_14
    Supported TIDL layer type --- Conv -- Conv_15
    Supported TIDL layer type --- Conv -- Conv_16
    Supported TIDL layer type --- Add -- Add_17
    Supported TIDL layer type --- Relu -- Relu_18
    Supported TIDL layer type --- Conv -- Conv_19
    Supported TIDL layer type --- Relu -- Relu_20
    Supported TIDL layer type --- Conv -- Conv_21
    Supported TIDL layer type --- Add -- Add_22
    Supported TIDL layer type --- Relu -- Relu_23
    Supported TIDL layer type --- Conv -- Conv_24
    Supported TIDL layer type --- Relu -- Relu_25
    Supported TIDL layer type --- Conv -- Conv_26
    Supported TIDL layer type --- Conv -- Conv_27
    Supported TIDL layer type --- Add -- Add_28
    Supported TIDL layer type --- Relu -- Relu_29
    Supported TIDL layer type --- Conv -- Conv_30
    Supported TIDL layer type --- Relu -- Relu_31
    Supported TIDL layer type --- Conv -- Conv_32
    Supported TIDL layer type --- Add -- Add_33
    Supported TIDL layer type --- Relu -- Relu_34
    Supported TIDL layer type --- Conv -- Conv_35
    Supported TIDL layer type --- Relu -- Relu_36
    Supported TIDL layer type --- Conv -- Conv_37
    Supported TIDL layer type --- Conv -- Conv_38
    Supported TIDL layer type --- Add -- Add_39
    Supported TIDL layer type --- Relu -- Relu_40
    Supported TIDL layer type --- Conv -- Conv_41
    Supported TIDL layer type --- Relu -- Relu_42
    Supported TIDL layer type --- Conv -- Conv_43
    Supported TIDL layer type --- Add -- Add_44
    Supported TIDL layer type --- Relu -- Relu_45
    Supported TIDL layer type --- Conv -- Conv_46
    Supported TIDL layer type --- Reshape -- Reshape_48
    Supported TIDL layer type --- Gemm -- Gemm_49
    Supported TIDL layer type --- Relu -- Relu_50
    Supported TIDL layer type --- Gemm -- Gemm_51
    Supported TIDL layer type --- Reshape -- Reshape_53

    Preliminary subgraphs created = 1
    Final number of subgraphs created are : 1, - Offloaded Nodes - 52, Total Nodes - 52
    SUGGESTION -- [TIDL_InnerProductLayer] Size larger than 2048 * 2048 is not optimal.
    Running runtimes graphviz - /home/leo/code/edgeai-tidl-tools-master/examples/osrt_python/ort/tidl_tools/tidl_graphVisualiser_runtimes.out ../../../model-artifacts//bestb//allowedNode.txt ../../../model-artifacts//bestb//tempDir/graphvizInfo.txt ../../../model-artifacts//bestb//tempDir/runtimes_visualization.svg
    *** In TIDL_createStateImportFunc ***
    Compute on node : TIDLExecutionProvider_TIDL_0_0
    0, Conv, 3, 1, input, 201
    1, Relu, 1, 1, 201, 129
    2, MaxPool, 1, 1, 129, 130
    3, Conv, 3, 1, 130, 204
    4, Relu, 1, 1, 204, 133
    5, Conv, 3, 1, 133, 207
    6, Add, 2, 1, 207, 136
    7, Relu, 1, 1, 136, 137
    8, Conv, 3, 1, 137, 210
    9, Relu, 1, 1, 210, 140
    10, Conv, 3, 1, 140, 213
    11, Add, 2, 1, 213, 143
    12, Relu, 1, 1, 143, 144
    13, Conv, 3, 1, 144, 216
    14, Relu, 1, 1, 216, 147
    15, Conv, 3, 1, 147, 219
    16, Conv, 3, 1, 144, 222
    17, Add, 2, 1, 219, 152
    18, Relu, 1, 1, 152, 153
    19, Conv, 3, 1, 153, 225
    20, Relu, 1, 1, 225, 156
    21, Conv, 3, 1, 156, 228
    22, Add, 2, 1, 228, 159
    23, Relu, 1, 1, 159, 160
    24, Conv, 3, 1, 160, 231
    25, Relu, 1, 1, 231, 163
    26, Conv, 3, 1, 163, 234
    27, Conv, 3, 1, 160, 237
    28, Add, 2, 1, 234, 168
    29, Relu, 1, 1, 168, 169
    30, Conv, 3, 1, 169, 240
    31, Relu, 1, 1, 240, 172
    32, Conv, 3, 1, 172, 243
    33, Add, 2, 1, 243, 175
    34, Relu, 1, 1, 175, 176
    35, Conv, 3, 1, 176, 246
    36, Relu, 1, 1, 246, 179
    37, Conv, 3, 1, 179, 249
    38, Conv, 3, 1, 176, 252
    39, Add, 2, 1, 249, 184
    40, Relu, 1, 1, 184, 185
    41, Conv, 3, 1, 185, 255
    42, Relu, 1, 1, 255, 188
    43, Conv, 3, 1, 188, 258
    44, Add, 2, 1, 258, 191
    45, Relu, 1, 1, 191, 192
    46, Conv, 3, 1, 192, 193
    47, Reshape, 2, 1, 193, 195
    48, Gemm, 3, 1, 195, 196
    49, Relu, 1, 1, 196, 197
    50, Gemm, 3, 1, 197, 198
    51, Reshape, 2, 1, 198, output

    Input tensor name - input
    Output tensor name - output
    In TIDL_onnxRtImportInit subgraph_name=output
    Layer 0, subgraph id output, name=output
    Layer 1, subgraph id output, name=input
    In TIDL_runtimesOptimizeNet: LayerIndex = 54, dataIndex = 53

    ************** Frame index 1 : Running float import *************
    In TIDL_runtimesPostProcessNet
    SUGGESTION: [TIDL_InnerProductLayer] Gemm_51 Size larger than 2048 * 2048 is not optimal.
    ****************************************************
    ** 1 WARNINGS 0 ERRORS **
    ****************************************************
    ************ in TIDL_subgraphRtCreate ************
    The soft limit is 2048
    The hard limit is 2048
    MEM: Init ... !!!
    MEM: Init ... Done !!!
    0.0s: VX_ZONE_INIT:Enabled
    0.5s: VX_ZONE_ERROR:Enabled
    0.6s: VX_ZONE_WARNING:Enabled
    0.1579s: VX_ZONE_INIT:[tivxInit:184] Initialization Done !!!
    ************ TIDL_subgraphRtCreate done ************
    ******* In TIDL_subgraphRtInvoke ********
    Layer, Layer Cycles,kernelOnlyCycles, coreLoopCycles,LayerSetupCycles,dmaPipeupCycles, dmaPipeDownCycles, PrefetchCycles,copyKerCoeffCycles,LayerDeinitCycles,LastBlockCycles, paddingTrigger, paddingWait,LayerWithoutPad,LayerHandleCopy, BackupCycles, RestoreCycles,

    1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    17, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    18, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    19, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    20, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    21, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    22, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    24, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    25, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    27, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    28, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    30, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    31, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    32, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    33, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    34, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    35, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    37, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    38, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    39, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    40, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,

    Sum of Layer Cycles 0
    Sub Graph Stats 237.000000 16919031.000000 186.000000
    ******* TIDL_subgraphRtInvoke done ********

    ********** Frame Index 1 : Running float inference **********
    ******* In TIDL_subgraphRtInvoke ********
    Layer, Layer Cycles,kernelOnlyCycles, coreLoopCycles,LayerSetupCycles,dmaPipeupCycles, dmaPipeDownCycles, PrefetchCycles,copyKerCoeffCycles,LayerDeinitCycles,LastBlockCycles, paddingTrigger, paddingWait,LayerWithoutPad,LayerHandleCopy, BackupCycles, RestoreCycles,

    1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    17, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    18, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    19, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    20, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    21, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    22, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    24, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    25, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    27, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    28, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    30, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    31, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    32, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    33, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    34, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    35, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    37, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    38, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    39, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    40, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,

    Sum of Layer Cycles 0
    Sub Graph Stats 319.000000 16811320.000000 131.000000
    ******* TIDL_subgraphRtInvoke done ********

    ********** Frame Index 2 : Running fixed point mode for calibration **********
    In TIDL_runtimesPostProcessNet

    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 16715.82 .... ..... ... .... .....
    # 1 . .. T 16788.76 .... ..... ... .... .....
    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 10349.83 .... ..... ... .... .....
    # 1 . .. T 10283.69 .... ..... ... .... .....


    ***************** Calibration iteration number 0 completed ************************

    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 10276.73 .... ..... ... .... .....
    # 1 . .. T 10274.64 .... ..... ... .... .....


    ***************** Calibration iteration number 1 completed ************************

    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 10418.43 .... ..... ... .... .....
    # 1 . .. T 10384.01 .... ..... ... .... .....


    ***************** Calibration iteration number 2 completed ************************

    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 10272.75 .... ..... ... .... .....
    # 1 . .. T 10359.20 .... ..... ... .... .....


    ***************** Calibration iteration number 3 completed ************************

    ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

    Processing config file #0 : /home/leo/code/edgeai-tidl-tools-master/model-artifacts/bestb/tempDir/output_tidl_io_.qunat_stats_config.txt
    Freeing memory for user provided Net
    ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

    # 0 . .. T 10398.03 .... ..... ... .... .....
    # 1 . .. T 10254.21 .... ..... ... .... .....


    ***************** Calibration iteration number 4 completed ************************

    ------------------ Network Compiler Traces -----------------------------
    NC running for device: 1
    Running with OTF buffer optimizations
    successful Memory allocation
    Rerunning network compiler for reshape

    ------------------ Network Compiler Traces -----------------------------
    NC running for device: 1
    Running with OTF buffer optimizations
    successful Memory allocation
    SUGGESTION: [TIDL_InnerProductLayer] Gemm_51 Size larger than 2048 * 2048 is not optimal.
    ****************************************************
    ** 1 WARNINGS 0 ERRORS **
    ****************************************************


    Completed_Model : 1, Name : bestb , Total time : 91370.38, Offload Time : 16865.18 , DDR RW MBs : 0, Output File : py_out_bestb_ppp.jpg


    ************ in TIDL_subgraphRtDelete ************
    MEM: Deinit ... !!!
    MEM: Alloc's: 26 alloc's of 743870520 bytes
    MEM: Free's : 26 free's of 743870520 bytes
    MEM: Open's : 0 allocs of 0 bytes
    MEM: Deinit ... Done !!!

  • 英文论坛贴下还是没有回复,可以帮忙问下吗

  • 抱歉,已经去催了,可能问题比较复杂,工程师需要花些时间回复。

  • 请看下面e2e工程师的回复。

    Please check out our extensive documentation on Performance and Accuracy here : 

    https://github.com/TexasInstruments/edgeai-tidl-tools/blob/master/docs/tidl_osr_debug.md

  • 我看了这个文档,里面涉及的参数我都是按照标准来的,准确率也是写的accuracy_level=1。板子上的检测效果确是和PC端出入很大。请问,什么原因导致了准确率的丢失?怎么解决?

  • 看到e2e帖子上您已经咨询工程师了,请关注帖子的回复。

  • 请看下面e2e工程师最新的回复。

    Can you share the basis for this comparison you doing ? 

    The model running on the board is quantized model, there is speed vs accuracy trade off that one need to consider while running the model on the target.

    If you are looking for accuracy tradeoff what parameter you specifically looking for such ?