yolov5_ultralytics_to_darknet.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.7.0+cu110'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# YOLO V5 Ultralytics model and config format to Darknet format \n",
    "# Adriano Santos.\n",
    "\n",
    "import torch\n",
    "import struct\n",
    "import yaml\n",
    "import os\n",
    "from configparser import ConfigParser\n",
    "\n",
    "# Input files\n",
    "pt_file_name = \"yolov5x.pt\"\n",
    "yaml_file_name = \"yolov5x.yaml\"\n",
    "\n",
    "# Output files\n",
    "cfg_file_name = \"yolov5x.cfg\"\n",
    "weights_file_name = \"yolov5x.weights\"\n",
    "\n",
    "\n",
    "if os.path.exists(cfg_file_name):\n",
    "    os.remove(cfg_file_name)\n",
    "    \n",
    "if os.path.exists(weights_file_name):\n",
    "    os.remove(weights_file_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "[net]\n",
      "\n",
      "width=416\n",
      "\n",
      "height=416\n",
      "\n",
      "channels=3\n",
      "\n",
      "nc=1\n",
      "\n",
      "depth_multiple=1.33\n",
      "\n",
      "width_multiple=1.25\n",
      "\n",
      "train=./romain/dataporctrain.txt\n",
      "\n",
      "val=./romain/dataporcvalidation.txt\n",
      "\n",
      "anchors=10, 13, 16, 30, 33, 23, 30, 61, 62, 45, 59, 119, 116, 90, 156, 198, 373, 326\n",
      "\n",
      "\n",
      "\n",
      "[Focus]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=64, 3\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=128, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=128\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=256, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=9\n",
      "\n",
      "args=256\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=512, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=9\n",
      "\n",
      "args=512\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1024, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[SPP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1024, [5, 9, 13]\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=1024, False\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=512, 1, 1\n",
      "\n",
      "\n",
      "\n",
      "[nn.Upsample]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args='None', 2, 'nearest'\n",
      "\n",
      "\n",
      "\n",
      "[Concat]\n",
      "\n",
      "from=-1, 6\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=512, False\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=256, 1, 1\n",
      "\n",
      "\n",
      "\n",
      "[nn.Upsample]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args='None', 2, 'nearest'\n",
      "\n",
      "\n",
      "\n",
      "[Concat]\n",
      "\n",
      "from=-1, 4\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=256, False\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=256, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[Concat]\n",
      "\n",
      "from=-1, 14\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=512, False\n",
      "\n",
      "\n",
      "\n",
      "[Conv]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=1\n",
      "\n",
      "args=512, 3, 2\n",
      "\n",
      "\n",
      "\n",
      "[Concat]\n",
      "\n",
      "from=-1, 10\n",
      "\n",
      "number=1\n",
      "\n",
      "args=1\n",
      "\n",
      "\n",
      "\n",
      "[BottleneckCSP]\n",
      "\n",
      "from=-1\n",
      "\n",
      "number=3\n",
      "\n",
      "args=1024, False\n",
      "\n",
      "\n",
      "\n",
      "[Detect]\n",
      "\n",
      "from=17, 20, 23\n",
      "\n",
      "number=1\n",
      "\n",
      "args='nc', 'anchors'\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Config file conversion (YAML to CFG)\n",
    "with open(cfg_file_name, 'w') as cfg:\n",
    "    with open(yaml_file_name,'r') as f:\n",
    "        f = yaml.load(f,Loader=yaml.FullLoader)\n",
    "        net = 0\n",
    "        for l in f:\n",
    "            if ((l != 'anchors')and(l !='backbone')and(l !='head')):\n",
    "                if net == 0:\n",
    "                    cfg.write('\\n[net]\\n')\n",
    "                    cfg.write(\"width={}\\n\".format(416))\n",
    "                    cfg.write(\"height={}\\n\".format(416))\n",
    "                    cfg.write(\"channels={}\\n\".format(3))\n",
    "                    net =1\n",
    "                cfg.write(\"{}={}\\n\".format(l,f[l]))\n",
    "            elif (l == 'anchors'):\n",
    "                anchors = []\n",
    "                for v in f[l]:\n",
    "                    anchors.extend(v)\n",
    "                cfg.write(\"anchors={}\\n\".format(str(anchors)[1:-1]))\n",
    "            elif((l == 'backbone') or (l =='head')): \n",
    "                for vector in f[l]:\n",
    "                    #net type\n",
    "                    cfg.write('\\n[{}]\\n'.format(vector[2]))\n",
    "                    #from\n",
    "                    if type(vector[0]) == int:\n",
    "                        cfg.write(\"from={}\\n\".format(str(vector[0])))\n",
    "                    elif type(vector[0]) == list:\n",
    "                        cfg.write(\"from={}\\n\".format(str(vector[0])[1:-1]))\n",
    "                    #number\n",
    "                    if type(vector[1]) == int:\n",
    "                        cfg.write(\"number={}\\n\".format(str(vector[1])))\n",
    "                    elif type(vector[1]) == list:\n",
    "                        cfg.write(\"number={}\\n\".format(str(vector[1])[1:-1]))\n",
    "                    #params\n",
    "                    if type(vector[3]) == int:\n",
    "                        cfg.write(\"args={}\\n\".format(str(vector[3])))\n",
    "                    elif type(vector[3]) == list:\n",
    "                        cfg.write(\"args={}\\n\".format(str(vector[3])[1:-1]))\n",
    "\n",
    "with open(cfg_file_name,'r') as f:\n",
    "    for l in f.readlines():\n",
    "        print(l)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model(\n",
      "  (model): Sequential(\n",
      "    (0): Focus(\n",
      "      (conv): Conv(\n",
      "        (conv): Conv2d(12, 80, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "    )\n",
      "    (1): Conv(\n",
      "      (conv): Conv2d(80, 160, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (2): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(160, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(160, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(80, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(80, 80, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(80, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (3): Conv(\n",
      "      (conv): Conv2d(160, 320, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (4): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(320, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(320, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (4): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (5): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (6): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (7): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (8): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (9): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (10): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (11): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (5): Conv(\n",
      "      (conv): Conv2d(320, 640, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (6): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (4): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (5): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (6): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (7): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (8): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (9): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (10): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (11): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (7): Conv(\n",
      "      (conv): Conv2d(640, 1280, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (8): SPP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv(\n",
      "        (conv): Conv2d(2560, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (m): ModuleList(\n",
      "        (0): MaxPool2d(kernel_size=5, stride=1, padding=2, dilation=1, ceil_mode=False)\n",
      "        (1): MaxPool2d(kernel_size=9, stride=1, padding=4, dilation=1, ceil_mode=False)\n",
      "        (2): MaxPool2d(kernel_size=13, stride=1, padding=6, dilation=1, ceil_mode=False)\n",
      "      )\n",
      "    )\n",
      "    (9): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(1280, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (10): Conv(\n",
      "      (conv): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (11): Upsample(scale_factor=2.0, mode=nearest)\n",
      "    (12): Concat()\n",
      "    (13): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(1280, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(1280, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (14): Conv(\n",
      "      (conv): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (15): Upsample(scale_factor=2.0, mode=nearest)\n",
      "    (16): Concat()\n",
      "    (17): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(640, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(640, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(160, 160, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(160, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (18): Conv(\n",
      "      (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (19): Concat()\n",
      "    (20): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(320, 320, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(320, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (21): Conv(\n",
      "      (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): Hardswish()\n",
      "    )\n",
      "    (22): Concat()\n",
      "    (23): BottleneckCSP(\n",
      "      (cv1): Conv(\n",
      "        (conv): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (cv2): Conv2d(1280, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv3): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (cv4): Conv(\n",
      "        (conv): Conv2d(1280, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "        (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "        (act): Hardswish()\n",
      "      )\n",
      "      (bn): BatchNorm2d(1280, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "      (act): LeakyReLU(negative_slope=0.1, inplace=True)\n",
      "      (m): Sequential(\n",
      "        (0): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (1): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (2): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "        (3): Bottleneck(\n",
      "          (cv1): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "          (cv2): Conv(\n",
      "            (conv): Conv2d(640, 640, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "            (bn): BatchNorm2d(640, eps=0.001, momentum=0.03, affine=True, track_running_stats=True)\n",
      "            (act): Hardswish()\n",
      "          )\n",
      "        )\n",
      "      )\n",
      "    )\n",
      "    (24): Detect(\n",
      "      (m): ModuleList(\n",
      "        (0): Conv2d(320, 18, kernel_size=(1, 1), stride=(1, 1))\n",
      "        (1): Conv2d(640, 18, kernel_size=(1, 1), stride=(1, 1))\n",
      "        (2): Conv2d(1280, 18, kernel_size=(1, 1), stride=(1, 1))\n",
      "      )\n",
      "    )\n",
      "  )\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "# Load model\n",
    "model = torch.load(pt_file_name, map_location=\"cpu\")['model'].float()  # load to FP32\n",
    "model.eval()\n",
    "\n",
    "# Model create\n",
    "with open(weights_file_name,'w') as f:\n",
    "    for k,v in model.state_dict().items():\n",
    "        list_data = list(v.reshape(-1).numpy())\n",
    "        f.write('{}'.format(k))\n",
    "        for value in list_data:\n",
    "            f.write(\" \")\n",
    "            f.write(struct.pack('>f',float(value)).hex())\n",
    "        f.write('\\n')\n",
    "        \n",
    "# Model informations\n",
    "print(model)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}