Add files using upload-large-folder tool

Other/Dataset/coco.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04462969ce7f8472f6b9d2f386148d0a5943cb73b31cdbd0271ca45f4fc6471f
+size 20251719680

Other/Dataset/imagenet.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dfb21fe16edc067351aa41b6de0c5b8aae0271dcf59363643e988ceb81201bf
+size 15541216050

Other/Dataset/mirflickr.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04e2a4c30166c60c6c7cbb1fc39128ef861d78d3d5ff1b5fc9921dee3b772e65
+size 3226849280

Other/Dataset/nuswide_v2_256.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2854210470fe7370885839236c32d19631eaaea6be6c98283fc6d631022f8a54
+size 12989818880

Other/Dataset/sop.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36ecb6d4465b55c22e72059c653c9ecb6f92fe3fcdb1b488bdf26c1db2107042
+size 3207342080

extract_delg_gldv2/embed_delg_gldv2_pretrained.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56f394c43b3fe43920d2d9e6fed5919595017ab194fe12b29870af2bd596773f
+size 16786945

extract_delg_gldv2/getTrainCode_gldv2-new-aug2022.ipynb

@@ -0,0 +1,341 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "c51e1ddd-f84d-4455-8923-8a71d965a3c7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import torch.nn as nn\n",
+    "import torch.nn.functional as F\n",
+    "import numpy as np\n",
+    "from scripts.train_helper import prepare_dataloader, prepare_model\n",
+    "from functions.metrics import calculate_mAP\n",
+    "from tqdm import tqdm\n",
+    "from collections import defaultdict\n",
+    "from utils.misc import AverageMeter, Timer\n",
+    "import configs\n",
+    "import os\n",
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "ba66f7f5-2d95-46b0-a2c8-a0bf03df7c99",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "device = torch.device(\"cuda:0\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "2651a4b8-5086-41e2-bbd7-44e0b4b5582d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "config = {\n",
+    "    'dataset': 'landmark',\n",
+    "    'dataset_kwargs': {\n",
+    "        \"crop\": 512,\n",
+    "        \"evaluation_protocol\": 1,\n",
+    "        \"extra_dataset\": 0,\n",
+    "        \"norm\": 2,\n",
+    "        \"remove_train_from_db\": False,\n",
+    "        \"reset\": False,\n",
+    "        \"resize\": 512,\n",
+    "        \"separate_multiclass\": False,\n",
+    "        \"use_db_as_train\": False,\n",
+    "        'use_random_augmentation': False\n",
+    "    },\n",
+    "    \"arch_kwargs\": {\n",
+    "        \"nclass\": 81313\n",
+    "    },\n",
+    "    'batch_size': 32\n",
+    "}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "f3529853-1d17-4e53-9d26-6c130290bc84",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "resnet = torch.load('/data/jiuntian/delg_pretrained/resnet50_delg_gldv2_pretrained_fixbn.pth', map_location=device)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "1de1b5dd-1ff4-47b3-9c43-3641a37f56c3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "embedding = torch.load('/data/jiuntian/delg_pretrained/embed_delg_gldv2_pretrained.pth', map_location=device)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "390525cc-85e0-4414-b9f7-782c4f58dbf9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "class GeM(nn.Module):\n",
+    "    \"\"\"Generalized Mean Pooling.\n",
+    "    Paper: https://arxiv.org/pdf/1711.02512.\n",
+    "    \"\"\"\n",
+    "\n",
+    "    def __init__(self, p: int = 3, eps: float = 1e-6):\n",
+    "        super().__init__()\n",
+    "        self.p = p\n",
+    "        self.eps = eps\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        return F.avg_pool2d(x.clamp(min=self.eps).pow(self.p), (x.size(-2), x.size(-1))).pow(1.0 / self.p)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "5d375bf8-21cb-4d98-a310-37e57b63c2fb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "gem = GeM(p=3)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "9bffe906-ad06-4f96-b2c1-3345bfb25b82",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "class ResNet50(nn.Module):\n",
+    "    def __init__(self):\n",
+    "        super(ResNet50, self).__init__()\n",
+    "\n",
+    "        model = resnet\n",
+    "        self.conv1 = model.conv1\n",
+    "        self.bn1 = model.bn1\n",
+    "        self.relu = model.relu\n",
+    "        self.maxpool = model.maxpool\n",
+    "        self.layer1 = model.layer1\n",
+    "        self.layer2 = model.layer2\n",
+    "        self.layer3 = model.layer3\n",
+    "        self.layer4 = model.layer4\n",
+    "#         self.avgpool = model.avgpool\n",
+    "\n",
+    "        self.embed = embedding\n",
+    "        self.gem = gem\n",
+    "\n",
+    "    def forward(self, x):\n",
+    "        x = self.conv1(x)\n",
+    "        x = self.bn1(x)\n",
+    "        x = self.relu(x)\n",
+    "        x = self.maxpool(x)\n",
+    "\n",
+    "        x = self.layer1(x)\n",
+    "        x = self.layer2(x)\n",
+    "        x = self.layer3(x)\n",
+    "        x = self.layer4(x)\n",
+    "\n",
+    "#         x = self.avgpool(x)\n",
+    "        x = self.gem(x)\n",
+    "    \n",
+    "        x = torch.flatten(x, 1)\n",
+    "        x = self.embed(x)\n",
+    "        return x"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "f0f6c671-7e7c-4abb-ad41-2f76e1ac4e07",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model = ResNet50()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "b1073f29-f58c-462a-a7f8-734ae0be03bb",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "model.to(device)\n",
+    "print()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "bfca6ca7-1d69-4a9a-a20d-8d777b6b3d6a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_dataset = configs.dataset(config, filename='train.txt', transform_mode='train', return_id=True)\n",
+    "\n",
+    "train_loader = configs.dataloader(train_dataset, config['batch_size'], shuffle=False, drop_last=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "d73beaa0-1090-40c9-a1a4-d7ac94a10f3c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_codes(model, test_loader, device, return_codes=True, return_id=True):\n",
+    "    model.eval()\n",
+    "    meters = defaultdict(AverageMeter)\n",
+    "    total_timer = Timer()\n",
+    "    batchtimer = Timer()\n",
+    "\n",
+    "    total_timer.tick()\n",
+    "\n",
+    "    ret_codes = []\n",
+    "    ret_id = []\n",
+    "    ret_labels = []\n",
+    "\n",
+    "    pbar = tqdm(test_loader, desc='Test', ascii=True, bar_format='{l_bar}{bar:10}{r_bar}')\n",
+    "    batchtimer.tick()\n",
+    "    for i, tuples in enumerate(pbar):\n",
+    "        (data, labels, image_id) = tuples\n",
+    "        with torch.no_grad():\n",
+    "            data = data.to(device)\n",
+    "            x = model(data)\n",
+    "\n",
+    "            ret_codes.append(x.cpu())\n",
+    "            ret_id.append(image_id)\n",
+    "            ret_labels.append(labels.cpu())\n",
+    "        batchtimer.toc()\n",
+    "        \n",
+    "        meters['time'].update(batchtimer.total)\n",
+    "        batchtimer.tick()\n",
+    "\n",
+    "        pbar.set_postfix({key: val.avg for key, val in meters.items()})\n",
+    "\n",
+    "    total_timer.toc()\n",
+    "    meters['total_time'].update(total_timer.total)\n",
+    "\n",
+    "    print(f'{\"; \".join([f\"{key}={val.avg:.4f}\" for key, val in meters.items()])}')\n",
+    "\n",
+    "    if return_codes:\n",
+    "        res = {\n",
+    "            'codes': torch.cat(ret_codes),\n",
+    "            'id': np.concatenate(ret_id) if len(ret_id) else np.array([]),\n",
+    "            'labels': torch.cat(ret_labels)\n",
+    "        }\n",
+    "        return meters, res\n",
+    "\n",
+    "    return meters"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "fed69362-0704-4cd7-b0bf-47e02f6dc672",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Test:   0%|          | 204/49390 [01:57<7:53:45,  1.73it/s, time=0.575] \n"
+     ]
+    },
+    {
+     "ename": "KeyboardInterrupt",
+     "evalue": "",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-17-d9be1cc8193f>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtrain_meter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_out\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_codes\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrain_loader\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreturn_codes\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreturn_id\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
+      "\u001b[0;32m<ipython-input-16-14943b1d03d0>\u001b[0m in \u001b[0;36mget_codes\u001b[0;34m(model, test_loader, device, return_codes, return_id)\u001b[0m\n\u001b[1;32m     19\u001b[0m             \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     20\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 21\u001b[0;31m             \u001b[0mret_codes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     22\u001b[0m             \u001b[0mret_id\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimage_id\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     23\u001b[0m             \u001b[0mret_labels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcpu\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
+     ]
+    }
+   ],
+   "source": [
+    "train_meter, train_out = get_codes(model, train_loader, device, return_codes=True, return_id=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "634152ff-79d3-4a7c-8df7-08d3c9593095",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "torch.save(train_out, 'delg_train_out.pth')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "e5f17bfe-a6e5-4e9b-a7f0-d3c325ef7993",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# torch.save(db_out, 'delg_db_out.pth')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "b039c7da-f53b-4130-9306-a69e68a370d9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# torch.save(test_out, 'delg_test_out.pth')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1beb6ffe-3a53-4eb7-b392-f5c619a43e32",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "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": "3.8.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

extract_delg_gldv2/resnet50_delg_gldv2_pretrained_fixbn.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a319d051ba704b22c5744c9d9fb0ab916be2c275f595cfe22c45ed77ecf4e867
+size 102566639

gldv2delgembed.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:840c995290db428ed5214b0d83dae7a94ad396cb3009a74975170b73d783427f
+size 17755228676

roxford5kdelgembed.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25fa05f0933b22524c2fb6fa9db0156e961a6824fcbec76fe833cbcacbeec03b
+size 83281920

rparis6kdelgembed.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69fa8bc36a7fbe2748c9296158b0dd0a2eeeda92444aa0fde1db26a7322bfba7
+size 52756480