量化后的onnx文件和量化后的bin文件使用hb_verifier 工具校验发现完全不一致
实际推理时也是如此，相同代码在docker中推理相同的img结果正常，在x5板端完全检测不出来，使用hb_verifier 工具查验后发现两个量化模型完全不一致，求大神帮忙看一下，不甚感激！！！
下面是hb_verifier 校验结果
root@OE-X5-CPU-1-2-6:/data/horizon_x5/data/ufld# hb_verifier -m model_output/ufld_288x800_bgr_quantized_model.onnx,model_output/ufld_288x800_bgr.bin -s True
/usr/local/lib/python3.10/dist-packages/paramiko/pkey.py
CryptographyDeprecationWarning: TripleDES has been moved to cryptography.hazmat.decrepit.ciphers.algorithms.TripleDES and will be removed from this module in 48.0.0.
“cipher”: algorithms.TripleDES,
/usr/local/lib/python3.10/dist-packages/paramiko/transport.py:259: CryptographyDeprecationWarning: TripleDES has been moved to cryptography.hazmat.decrepit.ciphers.algorithms.TripleDES and will be removed from this module in 48.0.0.
“class”: algorithms.TripleDES,
2025-05-08 15:53:42,450 INFO log will be stored in /data/horizon_x5/data/ufld/hb_verifier.log
2025-05-08 15:53:42,450 INFO HB_Verifier Starts…
2025-05-08 15:53:42,451 INFO verifier tool version 1.23.8
2025-05-08 15:53:42,451 INFO model: model_output/ufld_288x800_bgr_quantized_model.onnx,model_output/ufld_288x800_bgr.bin
2025-05-08 15:53:42,452 INFO board_ip: None
2025-05-08 15:53:42,452 INFO input: None
2025-05-08 15:53:42,452 INFO run_sim: True
2025-05-08 15:53:42,453 INFO dump_all_nodes_results: False
2025-05-08 15:53:42,454 INFO compare_digits: 5
2025-05-08 15:53:42,454 INFO =============== check params start ===============
1.23.10
2025-05-08 15:53:42,764 INFO host hrt version is 1.23.10
2025-05-08 15:53:59,635 INFO ================ check params end ================
2025-05-08 15:53:59,636 INFO ============== get input data start ==============
2025-05-08 15:54:00,032 INFO bin model input input shape: [1, 288, 800, 3]
2025-05-08 15:54:00,048 INFO bin input input data shape: (1, 288, 800, 3)
2025-05-08 15:54:08,099 INFO onnx input input shape: [1, 3, 288, 800]
2025-05-08 15:54:08,101 INFO onnx input input data shape: (1, 3, 288, 800)
2025-05-08 15:54:08,118 INFO =============== get input data end ===============
2025-05-08 15:54:08,118 INFO ================ Quanti infer log start =========================
2025-05-08 15:54:27,211 WARNING input[input] model input type is int8, input data type is uint8, will be convert.
2025-05-08 15:54:43,773 INFO ================= Quanti infer log end ==========================
2025-05-08 15:54:43,999 INFO ================== Sim infer log start ==========================
1.23.10
2025-05-08 15:55:02,620 INFO ================== Sim infer log end ==========================
2025-05-08 15:55:02,625 INFO ***************************************************************
2025-05-08 15:55:02,625 INFO compare source: Quanti.onnx VS Sim
2025-05-08 15:55:02,626 INFO compare model name: ufld_288x800_bgr_quantized_model VS ufld_288x800_bgr
Compare progress: 100%|###########################| 1/1 [00:00<00:00, 33.95it/s]
2025-05-08 15:55:02,660 INFO =============== Original output comparison results =================
2025-05-08 15:55:02,660 INFO Comparison results of original output is model_infer_output_0_output
2025-05-08 15:55:02,661 INFO mismatch result num: 14468
2025-05-08 15:55:02,663 INFO total result num: 14472
2025-05-08 15:55:02,663 INFO mismatch rate: 1.0
2025-05-08 15:55:02,664 INFO relative mismatch ratio: 1.0
2025-05-08 15:55:02,664 INFO max abs error: 0.5423800000000005
2025-05-08 15:55:02,665 WARNING raw output output and raw output output result Strict check FAILED
2025-05-08 15:55:02,666 WARNING Quanti.onnx and Sim result Strict check FAILED
下面是我的量化配置
# Copyright (c) 2020 Horizon Robotics.All Rights Reserved.
# The material in this file is confidential and contains trade secrets
# of Horizon Robotics Inc. This is proprietary information owned by
# Horizon Robotics Inc. No part of this work may be disclosed,
# reproduced, copied, transmitted, or used in any way for any purpose,
# without the express written permission of Horizon Robotics Inc.
# 模型转化相关的参数
# ------------------------------------
# model conversion related parameters
model_parameters:
# Onnx浮点网络数据模型文件
# -----------------------------------------------------------
# the model file of floating-point ONNX neural network data
onnx_model: ‘./ufld.onnx’
# 适用BPU架构
# --------------------------------
# the applicable BPU architecture
march: “bayes-e”
# 指定模型转换过程中是否输出各层的中间结果，如果为True，则输出所有层的中间输出结果，
# --------------------------------------------------------------------------------------
# specifies whether or not to dump the intermediate results of all layers in conversion
# if set to True, then the intermediate results of all layers shall be dumped
layer_out_dump: True
# 模型转换输出的结果的存放目录
# -----------------------------------------------------------
# the directory in which model conversion results are stored
working_dir: ‘model_output’
# 模型转换输出的用于上板执行的模型文件的名称前缀
# -----------------------------------------------------------------------------------------
# model conversion generated name prefix of those model files used for dev board execution
output_model_file_prefix: ‘ufld_288x800_bgr’
# remove_node_type: “Dequantize”
# 模型输入相关参数, 若输入多个节点, 则应使用’;'进行分隔, 使用默认缺省设置则写None
# --------------------------------------------------------------------------
# model input related parameters,
# please use “;” to seperate when inputting multiple nodes,
# please use None for default setting
input_parameters:
# (选填) 模型输入的节点名称, 此名称应与模型文件中的名称一致, 否则会报错, 不填则会使用模型文件中的节点名称
# --------------------------------------------------------------------------------------------------------
# (Optional) node name of model input,
# it shall be the same as the name of model file, otherwise an error will be reported,
# the node name of model file will be used when left blank
input_name: “”
# 网络实际执行时，输入给网络的数据格式，包括 nv12/rgb/bgr/yuv444/gray/featuremap,
# ------------------------------------------------------------------------------------------
# the data formats to be passed into neural network when actually performing neural network
# available options: nv12/rgb/bgr/yuv444/gray/featuremap,
input_type_rt: ‘bgr’
# 网络实际执行时输入的数据排布, 可选值为 NHWC/NCHW
# 若input_type_rt配置为nv12，则此处参数不需要配置
# ------------------------------------------------------------------
# the data layout formats to be passed into neural network when actually performing neural network, available options: NHWC/NCHW
# If input_type_rt is configured as nv12, then this parameter does not need to be configured
input_layout_rt: ‘NHWC’
# 网络训练时输入的数据格式，可选的值为rgb/bgr/gray/featuremap/yuv444
# --------------------------------------------------------------------
# the data formats in network training
# available options: rgb/bgr/gray/featuremap/yuv444
input_type_train: ‘rgb’
# 网络训练时输入的数据排布, 可选值为 NHWC/NCHW
# ------------------------------------------------------------------
# the data layout in network training, available options: NHWC/NCHW
input_layout_train: ‘NCHW’
# (选填) 模型网络的输入大小, 以’x’分隔, 不填则会使用模型文件中的网络输入大小，否则会覆盖模型文件中输入大小
# -------------------------------------------------------------------------------------------
# (Optional)the input size of model network, seperated by ‘x’
# note that the network input size of model file will be used if left blank
# otherwise it will overwrite the input size of model file
input_shape: ‘1x3x288x800’
# 网络实际执行时，输入给网络的batch_size, 默认值为1
# ---------------------------------------------------------------------
# the data batch_size to be passed into neural network when actually performing neural network, default value: 1
# input_batch: 1
# 网络输入的预处理方法，主要有以下几种：
# no_preprocess 不做任何操作
# data_mean 减去通道均值mean_value
# data_scale 对图像像素乘以data_scale系数
# data_mean_and_scale 减去通道均值后再乘以scale系数
# -------------------------------------------------------------------------------------------
# preprocessing methods of network input, available options:
# ‘no_preprocess’ indicates that no preprocess will be made
# ‘data_mean’ indicates that to minus the channel mean, i.e. mean_value
# ‘data_scale’ indicates that image pixels to multiply data_scale ratio
# ‘data_mean_and_scale’ indicates that to multiply scale ratio after channel mean is minused
norm_type: ‘data_mean_and_scale’
# 图像减去的均值, 如果是通道均值，value之间必须用空格分隔
# --------------------------------------------------------------------------
# the mean value minused by image
# note that values must be seperated by space if channel mean value is used
mean_value: 123.675 116.28 103.53
# 图像预处理缩放比例，如果是通道缩放比例，value之间必须用空格分隔
# ---------------------------------------------------------------------------
# scale value of image preprocess
# note that values must be seperated by space if channel scale value is used
scale_value: 0.0171248 0.017507 0.0174292
# 模型量化相关参数
# -----------------------------
# model calibration parameters
calibration_parameters:
# 模型量化的参考图像的存放目录，图片格式支持Jpeg、Bmp等格式，输入的图片
# 应该是使用的典型场景，一般是从测试集中选择20~100张图片，另外输入
# 的图片要覆盖典型场景，不要是偏僻场景，如过曝光、饱和、模糊、纯黑、纯白等图片
# 若有多个输入节点, 则应使用’;'进行分隔
# -------------------------------------------------------------------------------------------------
# the directory where reference images of model quantization are stored
# image formats include JPEG, BMP etc.
# should be classic application scenarios, usually 20~100 images are picked out from test datasets
# in addition, note that input images should cover typical scenarios
# and try to avoid those overexposed, oversaturated, vague,
# pure blank or pure white images
# use ‘;’ to seperate when there are multiple input nodes
cal_data_dir: ‘./calibration_data_rgb_f32’
# 校准数据二进制文件的数据存储类型，可选值为：float32, uint8
# calibration data binary file save type, available options: float32, uint8
cal_data_type: ‘float32’
# 如果输入的图片文件尺寸和模型训练的尺寸不一致时，并且preprocess_on为true，
# 则将采用默认预处理方法(skimage resize)，
# 将输入图片缩放或者裁减到指定尺寸，否则，需要用户提前把图片处理为训练时的尺寸
# ---------------------------------------------------------------------------------
# In case the size of input image file is different from that of in model training
# and that preprocess_on is set to True,
# shall the default preprocess method(skimage resize) be used
# i.e., to resize or crop input image into specified size
# otherwise user must keep image size as that of in training in advance
# preprocess_on: False
# 模型量化的算法类型，支持default、mix、kl、max、load，通常采用default即可满足要求
# 如不符合预期可先尝试修改为mix 仍不符合预期再尝试kl或max
# 当使用QAT导出模型时，此参数则应设置为load
# 相关参数的技术原理及说明请您参考用户手册中的PTQ原理及步骤中参数组详细介绍部分
# ----------------------------------------------------------------------------------
# The algorithm type of model quantization, support default, mix, kl, max, load, usually use default can meet the requirements.
# If it does not meet the expectation, you can try to change it to mix first. If there is still no expectation, try kl or max again.
# When using QAT to export the model, this parameter should be set to load.
# For more details of the parameters, please refer to the parameter details in PTQ Principle And Steps section of the user manual.
calibration_type: ‘mix’
# 编译器相关参数
# ----------------------------
# compiler related parameters
compiler_parameters:
# 编译策略，支持bandwidth和latency两种优化模式;
# bandwidth以优化ddr的访问带宽为目标；
# latency以优化推理时间为目标
# -------------------------------------------------------------------------------------------
# compilation strategy, there are 2 available optimization modes: ‘bandwidth’ and ‘lantency’
# the ‘bandwidth’ mode aims to optimize ddr access bandwidth
# while the ‘lantency’ mode aims to optimize inference duration
compile_mode: ‘latency’
# 设置debug为True将打开编译器的debug模式，能够输出性能仿真的相关信息，如帧率、DDR带宽占用等
# -----------------------------------------------------------------------------------
# the compiler’s debug mode will be enabled by setting to True
# this will dump performance simulation related information
# such as: frame rate, DDR bandwidth usage etc.
debug: True
# 编译模型指定核数，不指定默认编译单核模型, 若编译双核模型，将下边注释打开即可
# -------------------------------------------------------------------------------------
# specifies number of cores to be used in model compilation
# as default, single core is used as this value left blank
# please delete the "# " below to enable dual-core mode when compiling dual-core model
# core_num: 2
# 优化等级可选范围为O0~O3
# O0不做任何优化, 编译速度最快，优化程度最低,
# O1-O3随着优化等级提高，预期编译后的模型的执行速度会更快，但是所需编译时间也会变长。
# 推荐用O2做最快验证
# ----------------------------------------------------------------------------------------------------------
# optimization level ranges between O0~O3
# O0 indicates that no optimization will be made
# the faster the compilation, the lower optimization level will be
# O1-O3: as optimization levels increase gradually, model execution, after compilation, shall become faster
# while compilation will be prolonged
# it is recommended to use O2 for fastest verification
optimize_level: ‘O0’