eval_coco.py.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import cv2  # OpenCVライブラリ\n",
    "\n",
    "import matplotlib.pyplot as plt \n",
    "%matplotlib inline\n",
    "\n",
    "import torch\n",
    "\n",
    "import pickle\n",
    "\n",
    "import torch.utils.data as data\n",
    "from itertools import product as product\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.init as init\n",
    "import torch.nn.functional as F\n",
    "from torch.autograd import Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import dataset\n",
    "from utils.dataset import VOCDataset, COCODatasetTransform, make_datapath_list, Anno_xml2list, od_collate_fn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "model=\"efficientdet\"\n",
    "backbone = \"efficientnet-b0\"\n",
    "useBiFPN = True\n",
    "HALF = False # enable FP16\n",
    "DATASET = \"COCO\"\n",
    "retina = False # for trying retinanets\n",
    "scale = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "annotations\t\t  instances_trainval35k.json\r\n",
      "captions_train2014.json   instances_val2014.json\r\n",
      "captions_val2014.json\t  person_keypoints_train2014.json\r\n",
      "instances_train2014.json  person_keypoints_val2014.json\r\n"
     ]
    }
   ],
   "source": [
    "!ls ../data/coco/annotations/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loading annotations into memory...\n",
      "Done (t=11.46s)\n",
      "creating index...\n",
      "index created!\n",
      "loading annotations into memory...\n",
      "Done (t=6.25s)\n",
      "creating index...\n",
      "index created!\n"
     ]
    }
   ],
   "source": [
    "from dataset.coco import COCODetection\n",
    "import torch.utils.data as data\n",
    "from utils.dataset import VOCDataset, COCODatasetTransform, make_datapath_list, Anno_xml2list, od_collate_fn\n",
    "\n",
    "color_mean = (104, 117, 123)  # (BGR)の色の平均値\n",
    "if scale == 1:\n",
    "    input_size = 300  # 画像のinputサイズを300×300にする\n",
    "else:\n",
    "    input_size = 512\n",
    "\n",
    "## DatasetTransformを適応\n",
    "transform = COCODatasetTransform(input_size, color_mean)\n",
    "train_dataset = COCODetection(\"../data/coco/\", image_set=\"train2014\", phase=\"train\", transform=transform)\n",
    "val_dataset = COCODetection(\"../data/coco/\", image_set=\"val2014\", phase=\"val\", transform=transform)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = int(32/scale)\n",
    "\n",
    "train_dataloader = data.DataLoader(\n",
    "    train_dataset, batch_size=batch_size, shuffle=True, collate_fn=od_collate_fn, num_workers=8)\n",
    "\n",
    "val_dataloader = data.DataLoader(\n",
    "    val_dataset, batch_size=batch_size, shuffle=False, collate_fn=od_collate_fn, num_workers=8)\n",
    "\n",
    "# 辞書型変数にまとめる\n",
    "dataloaders_dict = {\"train\": train_dataloader, \"val\": val_dataloader}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "82"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "COCO_change_category = ['0','1','2','3','4','5','6','7','8','9','10','11','13','14','15','16','17','18','19','20',\n",
    "'21','22','23','24','25','26','27','28','31','32','33','34','35','36','37','38','39','40',\n",
    "'41','42','43','44','46','47','48','49','50','51','52','53','54','55','56','57','58','59',\n",
    "'60','61','62','63','64','65','67','70','72','73','74','75','76','77','78','79','80','81',\n",
    "'82','84','85','86','87','88','89','90']\n",
    "len(COCO_change_category)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded pretrained weights for efficientnet-b0\n",
      "EfficientNet(\n",
      "  (_conv_stem): Conv2dStaticSamePadding(\n",
      "    3, 32, kernel_size=(3, 3), stride=(2, 2), bias=False\n",
      "    (static_padding): ZeroPad2d(padding=(0, 1, 0, 1), value=0.0)\n",
      "  )\n",
      "  (_bn0): BatchNorm2d(32, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "  (_blocks): ModuleList(\n",
      "    (0): MBConvBlock(\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        32, 32, kernel_size=(3, 3), stride=[1, 1], groups=32, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 1, 1, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(32, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        32, 8, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        8, 32, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(16, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (1): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(96, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        96, 96, kernel_size=(3, 3), stride=[2, 2], groups=96, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(0, 1, 0, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(96, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        96, 4, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        4, 96, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(24, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (2): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(144, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        144, 144, kernel_size=(3, 3), stride=(1, 1), groups=144, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 1, 1, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(144, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        144, 6, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        6, 144, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(24, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (3): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(144, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        144, 144, kernel_size=(5, 5), stride=[2, 2], groups=144, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 2, 1, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(144, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        144, 6, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        6, 144, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(40, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (4): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(240, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        240, 240, kernel_size=(5, 5), stride=(1, 1), groups=240, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(240, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        240, 10, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        10, 240, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(40, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (5): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(240, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        240, 240, kernel_size=(3, 3), stride=[2, 2], groups=240, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(0, 1, 0, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(240, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        240, 10, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        10, 240, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(80, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (6): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        480, 480, kernel_size=(3, 3), stride=(1, 1), groups=480, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 1, 1, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        480, 20, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        20, 480, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(80, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (7): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        480, 480, kernel_size=(3, 3), stride=(1, 1), groups=480, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 1, 1, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        480, 20, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        20, 480, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(80, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (8): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        480, 480, kernel_size=(5, 5), stride=[1, 1], groups=480, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(480, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        480, 20, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        20, 480, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(112, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (9): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        672, 672, kernel_size=(5, 5), stride=(1, 1), groups=672, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        672, 28, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        28, 672, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(112, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (10): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        672, 672, kernel_size=(5, 5), stride=(1, 1), groups=672, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        672, 28, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        28, 672, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(112, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (11): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        672, 672, kernel_size=(5, 5), stride=[2, 2], groups=672, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 2, 1, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(672, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        672, 28, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        28, 672, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(192, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (12): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        1152, 1152, kernel_size=(5, 5), stride=(1, 1), groups=1152, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        1152, 48, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        48, 1152, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(192, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (13): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        1152, 1152, kernel_size=(5, 5), stride=(1, 1), groups=1152, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        1152, 48, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        48, 1152, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(192, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (14): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        1152, 1152, kernel_size=(5, 5), stride=(1, 1), groups=1152, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(2, 2, 2, 2), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        1152, 48, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        48, 1152, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(192, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "    (15): MBConvBlock(\n",
      "      (_expand_conv): Conv2dStaticSamePadding(\n",
      "        192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn0): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_depthwise_conv): Conv2dStaticSamePadding(\n",
      "        1152, 1152, kernel_size=(3, 3), stride=[1, 1], groups=1152, bias=False\n",
      "        (static_padding): ZeroPad2d(padding=(1, 1, 1, 1), value=0.0)\n",
      "      )\n",
      "      (_bn1): BatchNorm2d(1152, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_se_reduce): Conv2dStaticSamePadding(\n",
      "        1152, 48, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_se_expand): Conv2dStaticSamePadding(\n",
      "        48, 1152, kernel_size=(1, 1), stride=(1, 1)\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_project_conv): Conv2dStaticSamePadding(\n",
      "        1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "        (static_padding): Identity()\n",
      "      )\n",
      "      (_bn2): BatchNorm2d(320, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "      (_swish): MemoryEfficientSwish()\n",
      "    )\n",
      "  )\n",
      "  (_conv_head): Conv2dStaticSamePadding(\n",
      "    320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False\n",
      "    (static_padding): Identity()\n",
      "  )\n",
      "  (_bn1): BatchNorm2d(1280, eps=0.001, momentum=0.010000000000000009, affine=True, track_running_stats=True)\n",
      "  (_avg_pooling): AdaptiveAvgPool2d(output_size=1)\n",
      "  (_dropout): Dropout(p=0.2)\n",
      "  (_fc): Linear(in_features=1280, out_features=1000, bias=True)\n",
      "  (_swish): MemoryEfficientSwish()\n",
      ")\n",
      "320\n",
      "use BiFPN\n"
     ]
    }
   ],
   "source": [
    "from utils.efficientdet import EfficientDet\n",
    "if not DATASET == \"COCO\":\n",
    "    num_class = 21\n",
    "else:\n",
    "    num_class = 81\n",
    "\n",
    "if scale==1:\n",
    "    ssd_cfg = {\n",
    "        'num_classes': num_class,  # 背景クラスを含めた合計クラス数\n",
    "        'input_size': 300*scale,  # 画像の入力サイズ\n",
    "        'bbox_aspect_num': [4, 6, 6, 6, 4, 4],  # 出力するDBoxのアスペクト比の種類\n",
    "        'feature_maps': [37, 18, 9, 5, 3, 1],  # 各sourceの画像サイズ\n",
    "        'steps': [8, 16, 32, 64, 100, 300],  # DBOXの大きさを決める\n",
    "        'min_sizes': [30, 60, 111, 162, 213, 264],  # DBOXの大きさを決める\n",
    "        'max_sizes': [60, 111, 162, 213, 264, 315],  # DBOXの大きさを決める\n",
    "        'aspect_ratios': [[2], [2, 3], [2, 3], [2, 3], [2], [2]],\n",
    "    }\n",
    "elif scale==2:\n",
    "    ssd_cfg = {\n",
    "        'num_classes': num_class,  # 背景クラスを含めた合計クラス数\n",
    "        'input_size': 512,  # 画像の入力サイズ\n",
    "        'bbox_aspect_num': [4, 6, 6, 6, 4, 4],  # 出力するDBoxのアスペクト比の種類\n",
    "        'feature_maps': [64, 32, 16, 8, 4, 2],  # 各sourceの画像サイズ\n",
    "        'steps': [8, 16, 32, 64, 100, 300],  # DBOXの大きさを決める\n",
    "        'min_sizes': [30, 60, 111, 162, 213, 264]*scale,  # DBOXの大きさを決める\n",
    "        'max_sizes': [60, 111, 162, 213, 264, 315]*scale,  # DBOXの大きさを決める\n",
    "        'aspect_ratios': [[2], [2, 3], [2, 3], [2, 3], [2], [2]],\n",
    "    }\n",
    "\n",
    "# test if net works\n",
    "net = EfficientDet(phase=\"inference\", cfg=ssd_cfg, verbose=True, backbone=backbone, useBiFPN=useBiFPN)\n",
    "#out = net(torch.rand([1,3,input_size,input_size]))\n",
    "#print(out[0].size())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loaded the trained weights\n",
      "using: cuda:0\n"
     ]
    }
   ],
   "source": [
    "net_weights = torch.load('./weights/COCO_efficientnet-b0_600_BiFPN_50.pth',\n",
    "                         map_location={'cuda:0': 'cpu'})\n",
    "net.load_state_dict(net_weights)\n",
    "print('loaded the trained weights')\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(\"using:\", device)\n",
    "net = net.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_net(save_folder, net, cuda, testset, transform, thresh):\n",
    "    # dump predictions and assoc. ground truth to text file for now\n",
    "    filename = save_folder + 'result.txt'\n",
    "    num_images = len(testset)\n",
    "    for i in range(num_images):\n",
    "    print('Testing image {:d}/{:d}....'.format(i+1, num_images))\n",
    "    img = testset.pull_image(i)\n",
    "    x = torch.from_numpy(transform(img)[0]).permute(2, 0, 1)\n",
    "    x = Variable(x.unsqueeze(0))"
   ]
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}