海思NNIE开发·yolov7

官方预训练模型转换

1. 下载yolov7源码解压到本地，并配置基础运行环境。
2. 下载官方预训练模型

• yolov7-tiny.pt
• yolov7.pt
• …

3. 进入yolov7-main目录下，新建文件夹weights，并将步骤2中下载的权重文件放进去。
4. 修改models/yolo.py文件

    def forward(self, x):
        # x = x.copy()  # for profiling
        z = []  # inference output
        self.training |= self.export
        for i in range(self.nl):
            x[i] = self.m[i](x[i]).sigmoid()  # conv
        return x[0], x[1], x[2]

5. 新建export_nnie.py文件

import os
import torch
import onnx
from onnxsim import simplify
import onnxoptimizer
import argparse
from models.yolo import Detect, Model

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', type=str, default='./weights/yolov7.pt', help='initial weights path') 
    parser.add_argument('--cfg', type=str, default='./cfg/deploy/yolov7.yaml', help='initial weights path') 
    #================================================================
    opt = parser.parse_args()
    print(opt)

    #Save Only weights
    ckpt = torch.load(opt.weights, map_location=torch.device('cpu'))
    torch.save(ckpt['model'].state_dict(), opt.weights.replace(".pt", "-model.pt"))

    #Load model without postprocessing
    new_model = Model(opt.cfg)
    new_model.load_state_dict(torch.load(opt.weights.replace(".pt", "-model.pt"), map_location=torch.device('cpu')), False)
    new_model.eval()

    #save to JIT script
    example = torch.rand(1, 3, 640, 640)
    traced_script_module = torch.jit.trace(new_model, example)
    traced_script_module.save(opt.weights.replace(".pt", "-jit.pt"))

    #save to onnx
    f = opt.weights.replace(".pt", ".onnx")
    torch.onnx.export(new_model, example, f, verbose=False, opset_version=12,
                            training=torch.onnx.TrainingMode.EVAL,
                            do_constant_folding=True,
                            input_names=['data'],
                            output_names=['out0','out1','out2'])

    #onnxsim
    model_simp, check = simplify(f)
    assert check, "Simplified ONNX model could not be validated"
    onnx.save(model_simp, opt.weights.replace(".pt", "-sim.onnx"))

    #optimize onnx
    passes = ["extract_constant_to_initializer", "eliminate_unused_initializer"]
    optimized_model = onnxoptimizer.optimize(model_simp, passes)
    onnx.checker.check_model(optimized_model)
    onnx.save(optimized_model, opt.weights.replace(".pt", "-op.onnx"))
    print('finished exporting onnx')

6. 命令行执行python3 export_nnie.py脚本(默认为yolov7.pt, 加–weights参数可指定权重，–cfg指定模型配置文件)，转换成功会输出一下信息, 转换后的模型存于权重同级目录(*-op.onnx后缀模型)

Namespace(cfg='./cfg/deploy/yolov7.yaml', weights='./weights/yolov7.pt')
finished exporting onnx

7. 修改onnx网络结构，新建modify_onnx.py, yolov7-tiny不需要此步骤

import onnx
model = onnx.load("./weights/yolov7-op.onnx")
graph = model.graph

# print(onnx.helper.printable_graph(graph)) # 输出onnx的计算图

for index,n in enumerate(graph.node):
        
    if n.name in ["Add_291","Add_298","Add_305"]:
        graph.node.remove(n)
    
    if n.name == "Add_289":
        n.output[0] = "946"
    if n.name == "Add_296":
        n.output[0] = "955"
    if n.name == "Add_303":
        n.output[0] = "964"

for index,n in enumerate(graph.node):
    print(n.name)

onnx.checker.check_model(model)
onnx.save(model,"./weights/yolov7-op.onnx")

节点中Add(B=0)在转换caffe时会报错，直接删除节点, 并将下层输入替换为上层输出。

海思相机植入-模型转换篇

前期准备

• Hi3516/3519芯片系列软件定义相机
• Ruyistudio开发工具 用于转换NNIE模型
• vscode/MobaXterm 用于相机后台调试
• pycaffe环境 支持上采样、permute等自定义操作

详细流程

1. onnx->caffe
   
   可参考yolov5_onnx2caffe开源工程, 转换成功会输出一下信息。
   

2. 打开Ruyistudio开发工具，新建nnie项目, 主要soc version要与目标相机版本相匹配
   

3. 打开.cfg文件，配置onnx2nnie必要参数， 然后点击运行按钮
   

4. 运行成功会输出以下信息
   

海思相机植入-NNIE加载推理篇

• 使用海思NNIE服务接口加载推理模型
  
  

下集预告

• 海思NNIE开发·yolov8