瑞芯微板子使用探索【RK3588】

2023-11-04  本文已影响0人  georgeguo

瑞芯微板子使用探索【RK3588】

1 信息查看

1.1 查看RK系列

root@firefly:~# lspci
0002:20:00.0 PCI bridge: Fuzhou Rockchip Electronics Co., Ltd Device 3588 (rev 01)
0002:21:00.0 Network controller: Broadcom Inc. and subsidiaries Device 449d (rev 02

3588硬件信息:

NPU加速:

使用NPU需要下载RKNN SDK,RKNN SDK 为带有 NPU 的 RK3588S/RK3588 芯片平台提供编程接口,能够帮助用户部署使用 RKNN-Toolkit2 导出的 RKNN 模型,加速 AI 应用的落地。

1.2 查看npu的使用情况

cat /sys/kernel/debug/rknpu/load
watch -n 1 cat /sys/kernel/debug/rknpu/load

2 瑞芯微RKNN开发流程

B站视频教程: https://www.bilibili.com/video/BV1Kj411D78q?p=3&vd_source=9138e2a910cf9bbb083cd42a6750ed10

RKNN开发流程

上图中包含了RKNN模型开发中用到的所用项目:

① RKNN工具rknn-toolkit2https://github.com/rockchip-linux/rknn-toolkit2, RKNN-Toolkit-Lite2 provides Python programming interfaces for Rockchip NPU platform (RK3566, RK3568, RK3588, RK3588S) to help users deploy RKNN models and accelerate the implementation of AI applications.

② RKNN工具rknn-toolkit lite 2。这个工具不是独立的,包含在rknn-toolkit2中,对应目录为rknn-toolkit2/rknn_toolkit_lite2。提供Python版本的模型推理功能,只能推理,对应RKNNLite类。

③ rknpu2:项目地址 https://github.com/rockchip-linux/rknpu2.git, 该提供了RKNN 开发的C SDK,运行在边端。

④ rknn_server:盒子中的服务程序,用于接收连板运行的名。盒子启动后,默认启动。若连板推理时,发现版本不匹配,可以从RKNPU2中复制对应的动态库, 如:rknpu2/runtime/RK3588/Linux/librknn_api/aarch64/librknnrt.so。

3 模型转换

3.1 pth转ONNX

3.1.1 yolov5-det转onnx

yolov5官网提供了export.py脚本,环境配置好后,执行该脚本即可。

python3 export.py --weights ../models/yolov5s-det/13978/best.pt --include onnx --opset 12 --simplify

注意:RKNN必须使用opset=12

3.1.2 yolov8-det转onnx

参考:

构建yolov8的开发环境,因为涉及修改yolov8的部分代码,所以最好直接将yolov8代码下载下来,单独构建一个yolov8的运行环境;

yolov8官方封装了YOLO类,直接调用转换即可【当前使用的版本是 Ultralytics YOLOv8.0.200】

from ultralytics import YOLO

if __name__ == "__main__":

    src_pt_model = "../models/yolov8-det/13981/best.pt"
    dst_onnx_model = "../models/yolov8-det/13981/best.onnx"

    src_pt_model = "../models/yolov8-seg/13931/best.pt"
    dst_onnx_model = "../models/yolov8-seg/13931/best.onnx"

    # Load a model
    model = YOLO(src_pt_model)  # load a custom trained model

    # Export the model
    model.export(format='onnx')

3.2 ONNX转RKNN

3.2.1 yolov5 ONNX转RKNN【服务器端操作】

RKNN工具rknn-toolkit2:https://github.com/rockchip-linux/rknn-toolkit2, RKNN-Toolkit-Lite2 provides Python programming interfaces for Rockchip NPU platform (RK3566, RK3568, RK3588, RK3588S) to help users deploy RKNN models and accelerate the implementation of AI applications.

因为使用的是RK3588板子,所以安装rknn-toolkit2

pip3 install rknn_toolkit2==1.5.2 -i https://pypi.tuna.tsinghua.edu.cn/simple

直接使用pip3安装,提示ERROR: No matching distribution found for rknn_toolkit2==1.5.2,直接从github的release上下载,并安装,下载文件约356M,安装包为whl文件,解压后可以获得。包中仅仅支持36/38/310版本的python。

tar zxvf rknn-toolkit2-1.5.2.tar.gz
cd rknn-toolkit2-1.5.2/packages
pip3 install --no-dependencies rknn_toolkit2-1.5.2+b642f30c-cp38-cp38-linux_x86_64.whl -U -i https://pypi.tuna.tsinghua.edu.cn/simple   # 否则安装很多依赖包且报错,后续需要时再重新安装
pip3 install onnx onnxruntime onnxoptimizer onnxsim ruamel_yaml -i https://pypi.tuna.tsinghua.edu.cn/simple

如果直接在板子中安装,需要安装rknn-toolkit-lite 2。在rknn-toolkit2-1.5.2\rknn_toolkit_lite2\packages中找到rknn_toolkit_lite2-1.5.2-cp38-cp38-linux_aarch64.whl安装即可

参考:

环境搭建好之后,调用RKNN的API转换即可。下面的代码转换为yolov5 onnx转rknn示例。RKNN模型默认精度是fp16

import os
from rknn.api import RKNN

if __name__ == '__main__':

    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    src_onnx_model = "models/yolov5s-det/13978/best.onnx"
    dst_rknn_model = "models/yolov5s-det/13978/best_origin_3_output.rknn"

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(
        mean_values=[[0, 0, 0]],
        std_values=[[255, 255, 255]],
        target_platform='rk3588',
        quantized_method='channel',  # layer
        optimization_level=1  # 0  1  2  3
    )
    print('done')

    # Load ONNX model
    print('--> Loading model')
    ret = rknn.load_onnx(
        model=src_onnx_model,
        outputs=["/model.24/m.0/Conv_output_0", "/model.24/m.1/Conv_output_0", "/model.24/m.2/Conv_output_0"])
    if ret != 0:
        print('Load yolov5 failed!')
        exit(ret)
    print('done')

    # Build model
    print('--> Building model')
    ret = rknn.build(
        do_quantization=False,
        dataset=dataset,
        rknn_batch_size=1
    )
    if ret != 0:
        print('Build yolov5 failed!')
        exit(ret)
    print('done')

    # Export RKNN model
    print('--> Export RKNN model')
    ret = rknn.export_rknn(dst_rknn_model)
    if ret != 0:
        print('Export yolov5rknn failed!')
        exit(ret)
    print('done')

    ret = rknn.accuracy_analysis(inputs=[img_path])
    if ret != 0:
        print('Accuracy analysis failed!')
    print(ret)
    print('done')
    rknn.release()
/opt/data/code/yolov5-research/data/quantify_data/0/fa4f0905d4f1f68c07f2e0f3fc26ab39.jpg
/opt/data/code/yolov5-research/data/quantify_data/0/9dc8c98f67484b5fa921ee392fc68bc3.jpg
/opt/data/code/yolov5-research/data/quantify_data/0/8a047ffcc0c61a15632a54c73fe7f70e.jpg
/opt/data/code/yolov5-research/data/quantify_data/0/02f66416341f3224cb0123508ab98f1a.jpg
/opt/data/code/yolov5-research/data/quantify_data/0/eb6ac6aca0b14e9f766c7122d0d81fd1.jpg
/opt/data/code/yolov5-research/data/quantify_data/0/e4d231eb6464cd0f103bf72a06806264.jpg

3.2.2 yolov8 ONNX转RKNN【服务器端操作】

yolov8-det转rknn流程同yolov5。

4 模型量化

4.1 yolov5-det量化

4.1.1 普通量化

yolov5普通量化的示例代码如下,

import os
from rknn.api import RKNN

if __name__ == '__main__':

    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    src_onnx_model = "models/yolov5s-det/13978/best.onnx"
    dst_rknn_model = "models/yolov5s-det/13978/best_int8_3_output.rknn"

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(
        mean_values=[[0, 0, 0]],
        std_values=[[255, 255, 255]],
        target_platform='rk3588',
        quantized_dtype="asymmetric_quantized-8",
        # quantized_algorithm="mmse",  # normal
        quantized_method='channel',  # layer
        optimization_level=1  # 0  1  2  3
    )
    print('done')

    # Load ONNX model
    print('--> Loading model')
    ret = rknn.load_onnx(
        model=src_onnx_model,
        outputs=["/model.24/m.0/Conv_output_0", "/model.24/m.1/Conv_output_0", "/model.24/m.2/Conv_output_0"])
    if ret != 0:
        print('Load yolov5 failed!')
        exit(ret)
    print('done')

    # Build model
    print('--> Building model')
    ret = rknn.build(do_quantization=True, dataset=dataset, rknn_batch_size=1)
    if ret != 0:
        print('Build yolov5 failed!')
        exit(ret)
    print('done')

    # Export RKNN model
    print(f'--> Export RKNN model to {dst_rknn_model}')
    # ret = rknn.export_rknn(dst_rknn_model, cpp_gen_cfg=True, target='rk3588')
    ret = rknn.export_rknn(dst_rknn_model)
    if ret != 0:
        print('Export yolov5rknn failed!')
        exit(ret)
    print('done')

    ret = rknn.accuracy_analysis(inputs=[img_path], target='rk3588')
    if ret != 0:
        print('Accuracy analysis failed!')
    print(ret)
    print('done')

    rknn.release()

关键点:

yolov5_3_output.png

4.1.2 混合量化

混合量化就是对模型中进行int8量化之后精度损失较大的部分,再修改为fp16,不使用int8量化。包含两个步骤,具体可参考文档《Rockchip_User_Guide_RKNN_Toolkit2_CN-1.5.2.pdf》。

步骤1:hybrid_quantization_step1,yolov5-det示例代码

import os
from rknn.api import RKNN

if __name__ == "__main__":
    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    src_onnx_model = "models/yolov5s-det/13978/best.onnx"

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(
        mean_values=[[0, 0, 0]],
        std_values=[[255, 255, 255]],
        target_platform='rk3588'
    )
    print('done')

    # Load ONNX model
    print('--> Loading model')
    ret = rknn.load_onnx(
        model=src_onnx_model,
        outputs=["/model.24/m.0/Conv_output_0", "/model.24/m.1/Conv_output_0", "/model.24/m.2/Conv_output_0"])
    if ret != 0:
        print('Load yolov5 failed!')
        exit(ret)
    print('done')

    # Build model
    rknn.hybrid_quantization_step1(
        dataset=dataset,
        rknn_batch_size=1,
        proposal=True,
        proposal_dataset_size=1
    )

    rknn.release()

步骤2:hybrid_quantization_step2,使用步骤1的输出,作为步骤2 的输入。步骤1生成的best.quantization.cfg用于说明需要修回fp16的节点。若第一步骤中的proposal=False,这里必须手动修改。若proposal=True,则自动生成需要修改的节点。yolov5-det示例代码。

from rknn.api import RKNN

if __name__ == "__main__":
    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    dst_rknn_model = "models/yolov5s-det/13978/best_hybrid8_3_output.rknn"

    # Create RKNN object
    rknn = RKNN()

    # hybrid_quantization_step2, 输入均为第1步输出的文件
    rknn.hybrid_quantization_step2(
        model_input="best.model",
        data_input="best.data",
        model_quantization_cfg="best.quantization.cfg"
    )

    # Export RKNN model
    print('--> Export RKNN model')
    ret = rknn.export_rknn(dst_rknn_model, target='rk3588')
    if ret != 0:
        print('Export yolov5rknn failed!')
        exit(ret)
    print('done')

    rknn.accuracy_analysis(
        inputs=[img_path],
        target='rk3588',
    )


    rknn.release()

4.1.3 量化前后对比

注意:rknn模型也可以通过Netron进行查看,不过要使用较新的版本,下面是yolov5未量化、int8量化、混合精度的对比。

各种量化对比

4.1.4 量化过程遇到的问题

问题1:为何yolov5使用int8量化之后,推理速度反而变慢了?

量化前后,输出的维度不一样了,batch_size 变成了32。这个是因为我在服务器上量化时,把batch size设置为了32,把batch_size设置为1即可。

ret = rknn.build(do_quantization=True, dataset=dataset, rknn_batch_size=1)

问题2:按照yolov5的前处理进行时,推理结果异常

由于rknn默认使用的是data_format="nhwc",而正常yolov5使用的顺序是data_format="nchw",因此可以在前处理部分修改,也可以在inference调用是显示指定data_format="nchw"。但是在rk3588上使用data_format="nchw"时,提示不支持nchw,所以建议还是使用默认的“nhwc”顺序。

rk_result = self._rk_session.inference(inputs=[img], data_format="nhwc")

问题3:模型量化之后,后处理该如何做?

直接按照未量化前的数据格式输入模式,int8量化模型推理的置信度会全为0;

参考:

解决方法:yolov5 3个卷积之后有sigmoid操作,如果这个时候使用int8量化,则输出的结果都是0。解决方法加载onnx模型时提取卷积之后的结果,然后手动写后处理后处理操作。

# Load ONNX model
ret = rknn.load_onnx(
    model=src_onnx_model,
    outputs=["/model.24/m.0/Conv_output_0", "/model.24/m.1/Conv_output_0", "/model.24/m.2/Conv_output_0"])

问题4:yolov5 seg直接取output0和output1的输出进行后处理之后,可以获得正确的结果,但是取/model.22/Mul_2_output_0、/model.22/Split_output_1、/model.22/Concat_output_0、/model.22/proto/cv3/conv/Conv_output_0推理的结果存在截断的情况?
<img src="https://img.haomeiwen.com/i1700062/cc11052edfe858e1.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240" alt="轮廓被截断" style="zoom:50%;" />

原因分析,绘制出bbox,发现在后处理过程中,bbox缩放错误,导致了分割区域出现被切分的现象。

问题5:yolov5 seg直接转rknn,使用默认的2个输出,不进行量化,但是推理仍然无结果
原因分析:yolov5 seg训练时使用的参数是768x768,而使用export.py转onnx时,默认是640x640,虽然可以直接转rknn模型,但是推理无结果。解决方法是在导出脚本上添加模型的输入尺寸。

python3 export.py --weights ../models/yolov5s-seg/13933/best.pt --include onnx --opset 12 --simplify --img-size 768 768

4.2 yolov8-det量化

4.2.1 普通量化

不修改模型,导出默认的输出,由于sigmoid函数的影响,导致输出无结果,因此量化后保证模型可以使用,必须导出2个节点,或者6个节点。

导出两个节点的RKNN模型(无需修改模型代码直接导出即可):

import os
from rknn.api import RKNN

if __name__ == '__main__':

    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    src_onnx_model = "models/yolov8s-det/13980/best_normal.onnx"
    dst_rknn_model = "models/yolov8s-det/13980/best_2_output_int8.rknn"

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(
        mean_values=[[0, 0, 0]],
        std_values=[[255, 255, 255]],
        target_platform='rk3588',
        quantized_dtype="asymmetric_quantized-8",
        quantized_method='channel',  # layer
        optimization_level=1  # 0  1  2  3
    )
    print('done')

    # Load ONNX model
    print('--> Loading model')
    ret = rknn.load_onnx(
        model=src_onnx_model,
        outputs=['/model.22/Mul_2_output_0', '/model.22/Split_output_1']
    )
    if ret != 0:
        print('Load yolov8 failed!')
        exit(ret)
    print('done')

    # Build model
    print('--> Building model')
    ret = rknn.build(do_quantization=True, dataset=dataset, rknn_batch_size=1)
    if ret != 0:
        print('Build yolov8 failed!')
        exit(ret)
    print('done')

    # Export RKNN model
    print(f'--> Export RKNN model {dst_rknn_model}')
    ret = rknn.export_rknn(dst_rknn_model)
    if ret != 0:
        print('Export yolov8rknn failed!')
        exit(ret)
    print('done')

    ret = rknn.accuracy_analysis(inputs=[img_path])
    if ret != 0:
        print('Accuracy analysis failed!')
    print(ret)
    print('done')

导出6个输出的模型,需要先修改模型代码,再导出onnx,再转换。

修改1:ultralytics/nn/modules/head.py,在class Detect新增“导出onnx增加”下的代码。同时把forward中的函数替换成下面的forward。

class Detect(nn.Module):
    """YOLOv8 Detect head for detection models."""
    dynamic = False  # force grid reconstruction
    export = False  # export mode
    shape = None
    anchors = torch.empty(0)  # init
    strides = torch.empty(0)  # init

    # 导出onnx增加
    conv1x1 = nn.Conv2d(16, 1, 1, bias=False).requires_grad_(False)
    x = torch.arange(16, dtype=torch.float)
    conv1x1.weight.data[:] = nn.Parameter(x.view(1, 16, 1, 1))
    
    def forward(self, x):
        y = []
        for i in range(self.nl):
            t1 = self.cv2[i](x[i])
            t2 = self.cv3[i](x[i])
            y.append(self.conv1x1(t1.view(t1.shape[0], 4, 16, -1).transpose(2, 1).softmax(1)))
            # y.append(t2.sigmoid())
            y.append(t2)
        return y

修改2:ultralytics/engine/exporter.py。在函数export_onnx()将output_names替换掉,如下:

# output_names = ['output0', 'output1'] if isinstance(self.model, SegmentationModel) else ['output0']
output_names = ['reg1', 'cls1',  'reg2', 'cls2',  'reg3', 'cls3']

导出onnx

import os
from ultralytics import YOLO

if __name__ == "__main__":
    input_height, input_width = 640, 640
    src_pt_model = "models/yolov8s-det/13980/best.pt"

    # Load a model
    model = YOLO(src_pt_model)  # load a custom trained model

    # Export the model
    model.export(
        format='onnx',
        imgsz=[input_height, input_width],
        opset=12,
        verbose=True,
        simplify=True
    )

    src_onnx_model = "models/yolov8s-det/13980/best.onnx"
    dst_onnx_model = "models/yolov8s-det/13980/best_6_output.onnx"
    os.rename(src_onnx_model, dst_onnx_model)
    print(f"rename {src_onnx_model} to {dst_onnx_model}")

转换为6个输出的RKNN模型

import os
from rknn.api import RKNN


if __name__ == '__main__':

    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    src_onnx_model = "models/yolov8s-det/13980/best_6_output.onnx"
    dst_rknn_model = "models/yolov8s-det/13980/best_6_output_int8.rknn"

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(mean_values=[[0, 0, 0]], std_values=[[255, 255, 255]], target_platform='rk3588')
    print('done')

    # Load ONNX model
    print('--> Loading model')
    ret = rknn.load_onnx(
        model=src_onnx_model,
        outputs=['reg1', 'cls1',  'reg2', 'cls2',  'reg3', 'cls3']
    )
    if ret != 0:
        print('Load yolov8 failed!')
        exit(ret)
    print('done')

    # Build model
    print('--> Building model')
    ret = rknn.build(
        do_quantization=True,
        dataset=dataset,
        rknn_batch_size = 1
    )
    if ret != 0:
        print('Build yolov8 failed!')
        exit(ret)
    print('done')

    # Export RKNN model
    print(f'--> Export RKNN model to {dst_rknn_model}')
    ret = rknn.export_rknn(dst_rknn_model)
    if ret != 0:
        print('Export yolov8rknn failed!')
        exit(ret)
    print('done')

    ret = rknn.accuracy_analysis(inputs=[img_path], target="rk3588")
    if ret != 0:
        print('Accuracy analysis failed!')
    print(ret)
    print('done')

4.2.2 混合精度量化

混合量化步骤1:

import os
import shutil
from rknn.api import RKNN


if __name__ == "__main__":
    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"

    output_node_count = 2
    if output_node_count == 2:
        src_onnx_model = "models/yolov8s-det/13980/best_normal.onnx"
        tmp_model_dir = "models/yolov8s-det/13980/best_2_output_hybrid"
        os.makedirs(tmp_model_dir, exist_ok=True)
    elif output_node_count == 6:
        # 需要修改源码,实现best_6_output.onnx的导出
        src_onnx_model = "models/yolov8s-det/13980/best_6_output.onnx"
        tmp_model_dir = "models/yolov8s-det/13980/best_6_output_hybrid"
        os.makedirs(tmp_model_dir, exist_ok=True)

    # Create RKNN object
    rknn = RKNN()

    if not os.path.exists(src_onnx_model):
        print('model not exist')
        exit(-1)

    # pre-process config
    print('--> Config model')
    rknn.config(
        mean_values=[[0, 0, 0]],
        std_values=[[255, 255, 255]],
        target_platform='rk3588'
    )
    print('done')

    # Load ONNX model
    print('--> Loading model')
    if output_node_count == 2:
        ret = rknn.load_onnx(
            model=src_onnx_model,
            outputs=['/model.22/Mul_2_output_0', '/model.22/Split_output_1']
        )
        if ret != 0:
            print('Load yolov8 failed!')
            exit(ret)
    elif output_node_count == 6:
        ret = rknn.load_onnx(
            model=src_onnx_model,
            outputs=['reg1', 'cls1',  'reg2', 'cls2',  'reg3', 'cls3']
        )
        if ret != 0:
            print('Load yolov8 failed!')
            exit(ret)
        print('done')
    else:
        raise Exception(f"invalid output_node_count = {output_node_count}")
    print(f'output_node_count = {output_node_count}, done')

    # Build model
    rknn.hybrid_quantization_step1(
        dataset=dataset,
        rknn_batch_size=1,
        proposal=True,
        proposal_dataset_size=1
    )

    rknn.release()

    # move file
    file_prefix = os.path.basename(src_onnx_model)[0:-5]
    shutil.move(file_prefix + ".model", os.path.join(tmp_model_dir, file_prefix + ".model"))
    shutil.move(file_prefix + ".data", os.path.join(tmp_model_dir, file_prefix + ".data"))
    shutil.move(file_prefix + ".quantization.cfg", os.path.join(tmp_model_dir, file_prefix + ".quantization.cfg"))
    print(f"copy files to {tmp_model_dir}")

混合量化步骤2:

import os
from rknn.api import RKNN

if __name__ == "__main__":
    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"

    output_node_count = 2
    if output_node_count == 2:
        src_onnx_model = "models/yolov8s-det/13980/best_normal.onnx"
        tmp_model_dir = "models/yolov8s-det/13980/best_2_output_hybrid"
        dst_rknn_model = "models/yolov8s-det/13980/best_2_output_hybrid_custom.rknn"
        os.makedirs(tmp_model_dir, exist_ok=True)
    elif output_node_count == 6:
        # 需要修改源码,实现best_6_output.onnx的导出
        src_onnx_model = "models/yolov8s-det/13980/best_6_output.onnx"
        tmp_model_dir = "models/yolov8s-det/13980/best_6_output_hybrid"
        dst_rknn_model = "models/yolov8s-det/13980/best_6_output_hybrid.rknn"
        os.makedirs(tmp_model_dir, exist_ok=True)

    # Create RKNN object
    rknn = RKNN()

    # hybrid_quantization_step2, 输入均为第1步输出的文件
    file_prefix = os.path.basename(src_onnx_model)[0:-5]
    rknn.hybrid_quantization_step2(
        model_input=os.path.join(tmp_model_dir, file_prefix + ".model"),
        data_input=os.path.join(tmp_model_dir, file_prefix + ".data"),
        model_quantization_cfg= os.path.join(tmp_model_dir, file_prefix + ".quantization_custom.cfg")
    )

    # Export RKNN model
    print('--> Export RKNN model')
    ret = rknn.export_rknn(dst_rknn_model, target='rk3588')
    if ret != 0:
        print('Export yolov8rknn failed!')
        exit(ret)
    print('done')

    rknn.accuracy_analysis(
        inputs=[img_path],
        target='rk3588',
    )

    rknn.release()

4.2.3 量化过程中遇到的问题

问题1:yolov8导出onnx后,量化为int8之后,为什么置信度量化后全为0?

因为sigmoid的值域(0,1),int8量化后就为0了。所以去掉sigmoid。导出下面红色框中的两层即可,'/model.22/Mul_2_output_0', '/model.22/Split_output_1'。

yolov8去除的output

参考:

问题2:yolov8混合量化时报错?

E hybrid_quantization_step1: Catch exception when building RKNN model!
E hybrid_quantization_step1: Traceback (most recent call last):
E hybrid_quantization_step1:   File "rknn/api/rknn_base.py", line 2109, in rknn.api.rknn_base.RKNNBase.hybrid_quantization_step1
E hybrid_quantization_step1:   File "rknn/api/quantizer.py", line 265, in rknn.api.quantizer.Quantizer.save_hybrid_cfg
E hybrid_quantization_step1:   File "rknn/api/quantizer.py", line 268, in rknn.api.quantizer.Quantizer.save_hybrid_cfg
E hybrid_quantization_step1:   File "/opt/data/virtualenvs/yolov8/lib/python3.8/site-packages/ruamel/yaml/main.py", line 1229, in dump
E hybrid_quantization_step1:     error_deprecation('dump', 'dump', arg="typ='unsafe', pure=True")
E hybrid_quantization_step1:   File "/opt/data/virtualenvs/yolov8/lib/python3.8/site-packages/ruamel/yaml/main.py", line 1017, in error_deprecation
E hybrid_quantization_step1:     sys.exit(1)
E hybrid_quantization_step1: SystemExit: 1

参考:https://github.com/laitathei/YOLOv8-ONNX-RKNN-HORIZON-TensorRT-Segmentation/tree/master

ruamel_yaml版本的问题,默认安装的版本过高,导致执行失败,重新安装ruamel_yaml即可。

pip3 install ruamel_yaml==0.17.40 -i https://pypi.tuna.tsinghua.edu.cn/simple

问题3:yolov8 int8或者混合精度量化之后,无论使用1个输出、2个输出、6个输出,都无法检测结果,而fp16正常推理?

原因分析:出现该问题主要在预处理上,yolov8的预处理默认进行了归一化即除掉了255,导出时使用mean_values=[[0, 0, 0]], std_values=[[1, 1, 1]]若使用fp16时是没问题的,但int8时,先除255归一化之后,输入的值可能已经是空了,所以很难推理出结果。

解决方法,导出时使用mean_values=[[0, 0, 0]], std_values=[[255, 255, 255]],预处理中把除255去掉。说明,量化之后在前处理时,就不能再Normalize。

# image_data = np.array(img) / 255.0

问题4:yolov8 int8量化之后,2个输出时,检测框会发生偏移,6个输出时,检测框偏大,但不偏移。

原因分析:yolov8 2输出时, int8、fp16、hybrid转换之后,使用同一套前后处理代码,int8、hybrid之后的模型均存在检测框偏移的情况,从而可以推断是模型推理的问题。6个输出之所以不存在检测结果偏移的问题,是因为6个输出后都是基于fp32推理的。基于此,如果使用混合精度量化,模仿6输出,将6输出后的层全部设置为float16,应该就可以解决检测框偏移的问题。在best_6_output.quantization.cfg文件中添加一下内容,在生成混合精度的量化时,该问题解决。

custom_quantize_layers:
    /model.22/cv2.2/cv2.2.0/conv/Conv_output_0: float16
    /model.22/cv2.2/cv2.2.0/act/Mul_output_0: float16
    /model.22/cv2.2/cv2.2.1/conv/Conv_output_0: float16
    /model.22/cv2.2/cv2.2.1/act/Mul_output_0: float16
    /model.22/cv2.2/cv2.2.2/Conv_output_0: float16
    /model.22/cv3.2/cv3.2.0/conv/Conv_output_0: float16
    /model.22/cv3.2/cv3.2.0/act/Mul_output_0: float16
    /model.22/cv3.2/cv3.2.1/conv/Conv_output_0: float16
    /model.22/cv3.2/cv3.2.1/act/Mul_output_0: float16
    /model.22/cv3.2/cv3.2.2/Conv_output_0: float16
    /model.22/Concat_2_output_0: float16
    /model.22/Reshape_2_output_0_shape4_/model.22/Concat_3: float16
    /model.22/cv2.1/cv2.1.0/conv/Conv_output_0: float16
    /model.22/cv2.1/cv2.1.0/act/Mul_output_0: float16
    /model.22/cv2.1/cv2.1.1/conv/Conv_output_0: float16
    /model.22/cv2.1/cv2.1.1/act/Mul_output_0: float16
    /model.22/cv2.1/cv2.1.2/Conv_output_0: float16
    /model.22/cv3.1/cv3.1.0/conv/Conv_output_0: float16
    /model.22/cv3.1/cv3.1.0/act/Mul_output_0: float16
    /model.22/cv3.1/cv3.1.1/conv/Conv_output_0: float16
    /model.22/cv3.1/cv3.1.1/act/Mul_output_0: float16
    /model.22/cv3.1/cv3.1.2/Conv_output_0: float16
    /model.22/Concat_1_output_0: float16
    /model.22/Reshape_1_output_0_shape4_/model.22/Concat_3: float16
    /model.22/cv2.0/cv2.0.0/conv/Conv_output_0: float16
    /model.22/cv2.0/cv2.0.0/act/Mul_output_0: float16
    /model.22/cv2.0/cv2.0.1/conv/Conv_output_0: float16
    /model.22/cv2.0/cv2.0.1/act/Mul_output_0: float16
    /model.22/cv2.0/cv2.0.2/Conv_output_0: float16
    /model.22/cv3.0/cv3.0.0/conv/Conv_output_0: float16
    /model.22/cv3.0/cv3.0.0/act/Mul_output_0: float16
    /model.22/cv3.0/cv3.0.1/conv/Conv_output_0: float16
    /model.22/cv3.0/cv3.0.1/act/Mul_output_0: float16
    /model.22/cv3.0/cv3.0.2/Conv_output_0: float16
    /model.22/Concat_output_0: float16
    /model.22/Reshape_output_0_shape4_/model.22/Concat_3: float16
    /model.22/Concat_3_output_0: float16
    /model.22/Split_output_0_shape4: float16
    /model.22/Split_output_1_shape4: float16
    /model.22/dfl/Reshape_output_0: float16
    /model.22/dfl/Softmax_pre_tp: float16
    /model.22/dfl/Softmax_new: float16
    /model.22/dfl/conv/Conv_output_0: float16
    /model.22/dfl/Reshape_1_output_0_shape4_/model.22/Slice_2split: float16
    /model.22/dfl/Reshape_1_output_0_shape4_/model.22/Slice_2split_conv_/model.22/Slice_2split: float16
    /model.22/Slice_output_0_shape4_before_conv: float16
    /model.22/Slice_1_output_0: float16
    /model.22/Slice_output_0: float16
    /model.22/Sub_output_0: float16
    /model.22/Add_1_output_0: float16
    /model.22/Sub_1_output_0: float16
    /model.22/Concat_4_swap_concat_reshape_i1_out: float16
    /model.22/Add_2_output_0: float16
    /model.22/Div_1_output_0: float16
    /model.22/Concat_4_swap_concat_reshape_i0_out: float16
    /model.22/Concat_4_output_0_shape4: float16
    /model.22/Mul_2_output_0_shape4_before: float16
    /model.22/Mul_2_output_0: float16
    /model.22/Split_output_1: float16

说明量化之后模型的检测性能存在不稳定性,需要仔细分析,慎重量化。

4.2.4 yolov8-det量化总结

主要结论如下:

4.3 yolov8-seg量化

yolov8默认输出两个节点,分别是output0和output1,output0对应的是目标检测结果和部分分割结果,output1是部分分割结果。因此可以从output0中提取/model.22/Mul_2_output_0、/model.22/Split_output_1、/model.22/Concat_output_0,从output1的上一层提取/model.22/proto/cv3/conv/Conv_output_0,构成四个节点的输出,防止量化后,因sigmoid操作导致无检测结果。

5 模型部署

5.1 在板子上部署

参考:yolov5篇---yolov5训练pt模型并转换为rknn模型,部署在RK3588开发板上——从训练到部署全过程 https://blog.csdn.net/m0_46825740/article/details/128818516

下载rknpu2:https://github.com/rockchip-linux/rknpu2

下面的命令是在边端设备上使用官方提供的代码,编译完成后,进行推理的示例。

git clone https://github.com/rockchip-linux/rknpu2.git

- examples/rknn_yolov5_demo => #define OBJ_CLASS_NUM 3  # 修改为对应类别数
- coco_80_labels_list.txt   # 修改对应的标签

./build-linux_RK3588.sh
./rknn_yolov5_demo ./model/RK3588/yolov5s-640-640.rknn ./model/bus.jpg
./rknn_yolov5_video_demo ./model/RK3588/yolov5s-640-640.rknn 28ab8ba8a51a8e46eabc58575b6c208e.mp4 264
./rknn_yolov5_video_demo ./model/RK3588/yolov5s-640-640.rknn vlc-record-2023-09-22-11h14m49s.mp4 264

5.2 问题

问题1:固件中rknn_server版本较低
解决方法:更新rknn_server【开发模式】

cp /root/rknpu2/runtime/RK3588/Linux/rknn_server/aarch64/usr/bin/restart_rknn.sh /usr/bin
cp /root/rknpu2/runtime/RK3588/Linux/rknn_server/aarch64/usr/bin/start_rknn.sh /usr/bin
cp /root/rknpu2/runtime/RK3588/Linux/rknn_server/aarch64/usr/bin/rknn_server /usr/bin

问题2:复制更新rknn_server后,执行提示库的版本不对

root@firefly:/usr/bin# 1090443 RKNN SERVER loadRuntime(110): dlsym rknn_set_input_shapes failed: /lib/librknnrt.so: undefined symbol: rknn_set_input_shapes, reuqired librknnrt.so >= 1.5.2!
E RKNN: [07:43:34.070] 6, 4
E RKNN: [07:43:34.070] Invalid RKNN model version 6
E RKNN: [07:43:34.070] rknn_init, load model failed!
1090444 RKNN SERVER init(183): rknn_init fail! ret=-6
1090444 RKNN SERVER process_msg_init(381): Client 1 init model fail!

解决方法:复制对应的库,然后重启rknn_server即可

cp /root/rknpu2/runtime/RK3588/Linux/librknn_api/aarch64/librknnrt.so /lib/
restart_rknn.sh

问题3:E RKNN: [12:54:27.331] Mismatch driver version, librknnrt version: 1.5.2 (c6b7b351a@2023-08-23T15:28:22) requires driver version >= 0.7.0, but you have driver version: 0.6.4 which is incompatible!

原因及解决方法:出现这个错误,说明板子的固件版本较低,在官网下载对应的固件,并重新烧写即可。

6 模型推理

6.1 使用python rknn推理【服务器端】

加载转换生成的rknn模型进行推理。这里有个关键点,即rknn.init_runtime(target='rk3588', core_mask=RKNN.NPU_CORE_AUTO),若target设置为对应的目标设备型号,则进行连板推理,也就是通过rknn_server在边缘设备上推理。若target设置为None,则在模拟器上进行推理。

import os
import cv2
import tqdm
from rknn.api import RKNN
import multiprocessing

def process(rknn_model, img_path):
    # Create RKNN object
    rknn = RKNN()
    if not os.path.exists(rknn_model):
        print('model not exist')
        exit(-1)

    rknn.load_rknn(rknn_model)
    rknn.init_runtime(target='rk3588', core_mask=RKNN.NPU_CORE_AUTO)

    img = cv2.imread(filename=img_path)
    img = cv2.cvtColor(src=img, code=cv2.COLOR_BGR2RGB)
    img = cv2.resize(img, (640, 640))

    for item in tqdm.trange(10):
        result = rknn.inference(inputs=[img], data_format="nhwc")
        # print(result[0].shape)

    rknn.release()

def main():
    img_path = "data/car.jpg"
    dataset = "data/dataset.txt"
    rknn_model = "models/yolov5s-det/13978/best_hybrid8_3_output.rknn"
    p_count = 1
    p_list = list()
    for _ in range(p_count):
        p = multiprocessing.Process(target=process, args=(rknn_model, img_path))
        p_list.append(p)
    for p in p_list:
        p.start()
    for p in p_list:
        p.join()
    print("process over")

if __name__ == '__main__':
    main()

6.2 使用python rknnlite推理【边缘设备端】

参考rknn_toolkit_lite2\examples\inference_with_lite\test.py编写测试例子。

import os
import cv2
import tqdm
import time
import platform
import numpy as np
from rknnlite.api import RKNNLite

def get_host():
    device_compatible_node = '/proc/device-tree/compatible'
    # get platform and device type
    system = platform.system()
    machine = platform.machine()
    os_machine = system + '-' + machine
    if os_machine == 'Linux-aarch64':
        try:
            with open(device_compatible_node) as f:
                device_compatible_str = f.read()
                if 'rk3588' in device_compatible_str:
                    host = 'RK3588'
                else:
                    host = 'RK356x'
        except IOError:
            print('Read device node {} failed.'.format(device_compatible_node))
            exit(-1)
    else:
        host = os_machine
    return host

if __name__ == "__main__":
    host_name = get_host()
    print(host_name)

    rknn_file = "model/yolov5s-det/13978/best.rknn"
    rknn_file = "model/yolov5s-det/13978/best_int8.rknn"


    if not os.path.exists(rknn_file):
        print(f"{rknn_file} not exist")
        exit(0)

    rknn = RKNNLite(verbose=False, verbose_file="verbose.log")
    rknn.load_rknn(rknn_file)
    rknn.init_runtime(target=host_name, core_mask=RKNNLite.NPU_CORE_AUTO)

    ori_img = cv2.imread('./data/car.jpg')
    img = cv2.cvtColor(ori_img, cv2.COLOR_BGR2RGB)
    resized_image = cv2.resize(img, (640, 640))

    start_time = time.time()
    for i in tqdm.tqdm(range(100)):
        outputs = rknn.inference(inputs=[resized_image])
        print(len(outputs), outputs[0].shape)
    print(f"cost: {time.time() - start_time}, fps: {100 / (time.time() - start_time):.4f}")

    rknn.release()

6.3 问题

问题1:AttributeError: rknnlite/api/lib/hardware/DOLPHIN/linux-aarch64/librknn_api.so: undefined symbol: rknn_set_core_mask
解决方法:出现这个问题的原因是因为librknn_api.so的库是有问题的,直接将rknpu2库中的librknnrt.so 拷贝过去即可,执行命令如下:

# 备份
cp /opt/virtualenvs/rknn/lib/python3.8/site-packages/rknnlite/api/lib/hardware/DOLPHIN/linux-aarch64/librknn_api.so /opt/virtualenvs/rknn/lib/python3.8/site-packages/rknnlite/api/lib/hardware/DOLPHIN/linux-aarch64/librknn_api.so.bak

# 拷贝    
cp /root/rknpu2/runtime/RK3588/Linux/librknn_api/aarch64/librknnrt.so /opt/virtualenvs/rknn/lib/python3.8/site-packages/rknnlite/api/lib/hardware/DOLPHIN/linux-aarch64/librknn_api.so

问题2:W RKNN: [01:59:35.391] Output(output0):size_with_stride larger than model origin size, if need run OutputOperator in NPU, please call rknn_create_memory using size_with_stride.

这个告警暂时可以忽略,可以在代码中屏蔽告警。参考yolov5的前后处理:https://github.com/Applied-Deep-Learning-Lab/Yolov5_RK3588/blob/main/base/rk3588.py

from hide_warnings import hide_warnings
@hide_warnings
def rk_infer(self, img):
    rk_result = self._rk_session.inference(inputs=[img], data_format="nhwc")
    return rk_result

问题3:如何选中使用的NPU,即如何设置core_mask?

self._rk_session.init_runtime(target=self.host_name, core_mask=RKNNLite.NPU_CORE_AUTO)
self._rk_session.init_runtime(target=self.host_name, core_mask=RKNNLite.NPU_CORE_0_1_2)

解决方法:通过RKNNLite.NPU_CORE_AUTO/RKNNLite.NPU_CORE_0_1_2/RKNNLite.NPU_CORE_0都可以设置使用那个NPU,测试发现NPU_CORE_AUTO NPU的利用率最高。

7 其他

A1 镜像烧录步骤

参考
-【ROC-RK3568-PC开发板试用体验】烧录Ubuntu20.04系统:https://dev.t-firefly.com/thread-124315-1-1.html

步骤1:安装驱动

注意若通过linux连接到板子时,需要安装驱动adb

apt install adb

步骤2.连接设备

步骤3:整体固件【在官网下载对应的固件https://www.t-firefly.com/doc/download/183.html

1)选择整体固件方式:解压RK官方烧录工具RKDevTool_Release_v2.92.zip,打开RKDevTool.exe,会自动识别到ADB设备;
2)按【切换】按钮,进入loader烧录模式;
3)按【固件】按钮,选择要升级的固件文件,加载固件之后,然后点击【升级】按钮,等待烧写为完成即可

步骤4:镜像烧录完之后,安装必要的软件

apt install git cmake g++ -y 
apt install lrzsz tree vim htop -y 

A2 RK3588运行Docker

检测板子中Docker的运行环境

git clone https://github.com/moby/moby.git
cd moby/contrib
./check-config.sh .config

参考:https://blog.csdn.net/xiaoning132/article/details/130541520

板子中安装Docker

## 卸载旧版本
sudo apt-get remove docker docker-engine docker.io containerd runc

## 设置存储库
sudo apt-get update -y
sudo apt-get install -y ca-certificates curl gnupg lsb-release

sudo mkdir -p /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg

echo \
"deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
$(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

## 安装 Docker
sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-compose-plugin

## 验证是否安装成功

sudo docker run hello-world

参考:https://github.com/DHDAXCW/Rk3588-Docker

A3 GStreamer拉流

参考:

安装gstreamer


apt-get update

apt-get install libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev libgstreamer-plugins-bad1.0-dev gstreamer1.0-plugins-base gstreamer1.0-plugins-good gstreamer1.0-plugins-bad gstreamer1.0-plugins-ugly gstreamer1.0-libav gstreamer1.0-doc gstreamer1.0-tools gstreamer1.0-x gstreamer1.0-alsa gstreamer1.0-gl gstreamer1.0-gtk3 gstreamer1.0-qt5 gstreamer1.0-pulseaudio

apt-get install libunwind8-dev

apt-get install libgtk2.0-dev pkg-config

# 查看安装结果
dpkg -l | grep gstreamer

使用命令测试gstreamer

# hello world
gst-launch-1.0 videotestsrc ! videoconvert ! autovideosink

# Adding a capability to the pipeline
gst-launch-1.0 videotestsrc ! video/x-raw, format=BGR ! autovideoconvert ! ximagesink

# Setting width, height and framerate
gst-launch-1.0 videotestsrc ! video/x-raw, format=BGR ! autovideoconvert ! videoconvert ! video/x-raw, width=640, height=480, framerate=1/2 ! ximagesink

# rtsp
gst-launch-1.0 rtspsrc location=rtsp://172.18.18.202:5554/T_Perimeter_ball001 latency=10 ! queue ! rtph264depay ! h264parse ! avdec_h264 ! videoconvert ! videoscale ! video/x-raw,width=640,height=480 ! ximagesink

使用opencv测试gstreamner

步骤1:重新编译opencv。opencv默认未开启gstreamer的支持,所以需要先重新编译opencv

git clone https://github.com/opencv/opencv.git
cd opencv 
mkdir build && cd build


cmake -D WITH_GSTREAMER=ON \
-D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D BUILD_opencv_python2=OFF \
-D BUILD_opencv_python3=ON \
-D PYTHON3_PACKAGES_PATH=/usr/local/lib/python3.8/dist-packages/ \
-D PYTHON3_LIBRARY=/usr/lib/python3.8/config-3.8-aarch64-linux-gnu/libpython3.8.so \
-D OPENCV_GENERATE_PKGCONFIG=YES ..


make -j6 && make install

cd /etc/ld.so.conf.d/ # 切换目录
touch opencv.conf # 新建opencv配置文件
echo /usr/local/lib/ > opencv.conf # 填写opencv编译后库所在的路径
sudo ldconfig

步骤2:使用下面脚本测试gstreamer

# opencv【需要重新编译opencv并启用WITH_GSTREAMER=ON】

import cv2
gstreamer_str = "sudo gst-launch-1.0 rtspsrc location=rtsp://172.18.18.202:5554/T_Perimeter_ball001 latency=1000 ! queue ! rtph264depay ! h264parse ! avdec_h264 ! videoconvert ! videoscale ! video/x-raw,width=640,height=480,format=BGR ! appsink drop=1"
cap = cv2.VideoCapture(gstreamer_str, cv2.CAP_GSTREAMER)
print(cap.isOpened())
while(cap.isOpened()):
    ret, frame = cap.read()
    if ret:
        cv2.imshow("Input via Gstreamer", frame)
        if cv2.waitKey(25) & 0xFF == ord('q'):
            break

cap.release()
cv2.destroyAllWindows()

遇到的问题

问题1:未安装libgtk2.0-dev导致

Traceback (most recent call last):
  File "opencv_demo.py", line 14, in <module>
    cv2.destroyAllWindows()
cv2.error: OpenCV(4.8.0-dev) /root/code/opencv/modules/highgui/src/window.cpp:1266: error: (-2:Unspecified error) The function is not implemented. Rebuild the library with Windows, GTK+ 2.x or Cocoa support. If you are on Ubuntu or Debian, install libgtk2.0-dev and pkg-config, then re-run cmake or configure script in function 'cvDestroyAllWindows'

问题2:无法运行gstream,camke之后GStreamer显示为No

Video I/O:
  DC1394:                      YES (2.2.5)
  FFMPEG:                      YES
    avcodec:                   YES (58.54.100)
    avformat:                  YES (58.29.100)
    avutil:                    YES (56.31.100)
    swscale:                   YES (5.5.100)
    avresample:                YES (4.0.0)
  GStreamer:                   NO
  v4l/v4l2:                    YES (linux/videodev2.h)

原因:没有安装apt-get install libunwind8-dev,导致的。可以使用下面的代码查看是否支持gstreamer

import cv2
print(cv2.getBuildInformation())

问题3:fail to load module gail

解决方法:出现这个问题的原因是因为librknn_api

sudo apt-get install libgail-common

GStreamer Python:https://github.com/GStreamer/gst-python
https://gist.github.com/liviaerxin/9934a5780f5d3fe5402d5986fc32d070

git clone https://github.com/GStreamer/gst-python.git
cd gst-python

GSTREAMER_VERSION=$(gst-launch-1.0 --version | grep version | tr -s ' ' '\n' | tail -1)
git checkout $GSTREAMER_VERSION

PYTHON=/usr/bin/python3.8
LIBPYTHON=$($PYTHON -c 'from distutils import sysconfig; print(sysconfig.get_config_var("LDLIBRARY"))')
LIBPYTHONPATH=$(dirname $(ldconfig -p | grep -w $LIBPYTHON | head -1 | tr ' ' '\n' | grep /))
PREFIX=$(dirname $(dirname $(which python3))) # in jetson nano, `PREFIX=~/.local` to use local site-packages,

LIBPYTHON=libpython3.8.so
LIBPYTHONPATH=/lib/aarch64-linux-gnu
PREFIX=/usr

./autogen.sh --disable-gtk-doc --noconfigure
./configure --with-libpython-dir=$LIBPYTHONPATH --prefix $PREFIX

make
make install

错误:No package 'pygobject-3.0' found

apt install -y python-gi-dev

错误:checking for headers required to compile python extensions... not found
configure: error: could not find Python headers

sudo apt-get install python3-dev libpython3-dev

checking for PYGOBJECT... yes
checking for libraries required to embed python... no
configure: error: Python libs not found. Windows requires Python modules to be explicitly linked to libpython
上一篇下一篇

猜你喜欢

热点阅读