{ "cells": [ { "cell_type": "markdown", "id": "a2103c83", "metadata": { "origin_pos": 0 }, "source": [ "# Phân đoạn ngữ nghĩa và tập dữ liệu\n", ":label:`sec_semantic_segmentation`\n", "\n", "Khi thảo luận về các tác vụ phát hiện đối tượng trong :numref:`sec_bbox`—:numref:`sec_rcnn`, các hộp giới hạn hình chữ nhật được sử dụng để dán nhãn và dự đoán các đối tượng trong ảnh. Phần này sẽ thảo luận về vấn đề phân đoạn *ngữ nghĩa *, tập trung vào cách chia hình ảnh thành các vùng thuộc các lớp ngữ nghĩa khác nhau. Khác với phát hiện đối tượng, phân đoạn ngữ nghĩa nhận ra và hiểu những gì có trong hình ảnh ở cấp độ pixel: ghi nhãn và dự đoán các vùng ngữ nghĩa của nó ở mức pixel. :numref:`fig_segmentation` hiển thị nhãn của chó, mèo và nền của hình ảnh trong phân khúc ngữ nghĩa. So với trong phát hiện đối tượng, các đường viền cấp pixel được dán nhãn trong phân đoạn ngữ nghĩa rõ ràng là hạt mịn hơn. \n", "\n", "![Labels of the dog, cat, and background of the image in semantic segmentation.](../img/segmentation.svg)\n", ":label:`fig_segmentation`\n", "\n", "## Phân đoạn hình ảnh và Phân đoạn phiên bản\n", "\n", "Ngoài ra còn có hai nhiệm vụ quan trọng trong lĩnh vực thị giác máy tính tương tự như phân đoạn ngữ nghĩa, đó là phân đoạn hình ảnh và phân đoạn phiên bản. Chúng tôi sẽ phân biệt ngắn gọn chúng với phân khúc ngữ nghĩa như sau. \n", "\n", "* *Phân khúc hình ảnh* chia một hình ảnh thành nhiều vùng cấu thành. Các phương pháp cho loại vấn đề này thường sử dụng mối tương quan giữa các pixel trong hình ảnh. Nó không cần thông tin nhãn về pixel hình ảnh trong quá trình đào tạo và nó không thể đảm bảo rằng các vùng được phân đoạn sẽ có ngữ nghĩa mà chúng tôi hy vọng sẽ có được trong quá trình dự đoán. Chụp ảnh trong :numref:`fig_segmentation` làm đầu vào, phân đoạn hình ảnh có thể chia chó thành hai vùng: một vùng che miệng và mắt chủ yếu là màu đen, và phần còn lại bao phủ phần còn lại của cơ thể chủ yếu là màu vàng.\n", "* * Phân khúc Instance* còn được gọi là * phát hiện và phân đoạn đồng thời*. Nó nghiên cứu làm thế nào để nhận ra các vùng cấp pixel của mỗi đối tượng trong một hình ảnh. Khác với phân đoạn ngữ nghĩa, phân đoạn phiên bản cần phân biệt không chỉ ngữ nghĩa, mà còn các trường hợp đối tượng khác nhau. Ví dụ, nếu có hai chú chó trong hình ảnh, phân đoạn ví dụ cần phân biệt một pixel thuộc về cái nào trong hai chú chó.\n", "\n", "## Tập dữ liệu phân đoạn ngữ nghĩa Pascal VOC2012\n", "\n", "[**Trên tập dữ liệu phân đoạn ngữ nghĩa quan trọng nhất là [Pascal VOC2012](http://host.robots.ox.ac.uk/pascal/VOC/voc2012/).**] Sau đây, chúng ta sẽ xem tập dữ liệu này.\n" ] }, { "cell_type": "code", "execution_count": 1, "id": "934de491", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:31:50.252622Z", "iopub.status.busy": "2022-04-26T12:31:50.252219Z", "iopub.status.idle": "2022-04-26T12:31:52.388482Z", "shell.execute_reply": "2022-04-26T12:31:52.387513Z" }, "origin_pos": 2, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "%matplotlib inline\n", "import os\n", "import torch\n", "import torchvision\n", "from d2l import torch as d2l" ] }, { "cell_type": "markdown", "id": "e4c5f2fb", "metadata": { "origin_pos": 3 }, "source": [ "Tệp tar của tập dữ liệu khoảng 2 GB, vì vậy có thể mất một thời gian để tải xuống tệp. Tập dữ liệu được trích xuất nằm ở `../data/VOCdevkit/VOC2012`.\n" ] }, { "cell_type": "code", "execution_count": 2, "id": "139637ba", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:31:52.392483Z", "iopub.status.busy": "2022-04-26T12:31:52.391961Z", "iopub.status.idle": "2022-04-26T12:32:52.929476Z", "shell.execute_reply": "2022-04-26T12:32:52.928701Z" }, "origin_pos": 4, "tab": [ "pytorch" ] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Downloading ../data/VOCtrainval_11-May-2012.tar from http://d2l-data.s3-accelerate.amazonaws.com/VOCtrainval_11-May-2012.tar...\n" ] } ], "source": [ "#@save\n", "d2l.DATA_HUB['voc2012'] = (d2l.DATA_URL + 'VOCtrainval_11-May-2012.tar',\n", " '4e443f8a2eca6b1dac8a6c57641b67dd40621a49')\n", "\n", "voc_dir = d2l.download_extract('voc2012', 'VOCdevkit/VOC2012')" ] }, { "cell_type": "markdown", "id": "189b3f3c", "metadata": { "origin_pos": 5 }, "source": [ "Sau khi nhập đường dẫn `../data/VOCdevkit/VOC2012`, chúng ta có thể thấy các thành phần khác nhau của tập dữ liệu. Đường dẫn `ImageSets/Segmentation` chứa các tệp văn bản chỉ định các mẫu đào tạo và thử nghiệm, trong khi các đường dẫn `JPEGImages` và `SegmentationClass` lưu trữ hình ảnh đầu vào và nhãn cho mỗi ví dụ tương ứng. Nhãn ở đây cũng ở định dạng hình ảnh, có cùng kích thước với hình ảnh đầu vào được dán nhãn của nó. Bên cạnh đó, các pixel có cùng màu trong bất kỳ hình ảnh nhãn nào thuộc cùng một lớp ngữ nghĩa. Sau đây xác định hàm `read_voc_images` thành [** đọc tất cả các hình ảnh đầu vào và nhãn vào bộ nhớ**].\n" ] }, { "cell_type": "code", "execution_count": 3, "id": "67987c19", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:52.935238Z", "iopub.status.busy": "2022-04-26T12:32:52.935003Z", "iopub.status.idle": "2022-04-26T12:32:58.604317Z", "shell.execute_reply": "2022-04-26T12:32:58.603491Z" }, "origin_pos": 7, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "def read_voc_images(voc_dir, is_train=True):\n", " \"\"\"Read all VOC feature and label images.\"\"\"\n", " txt_fname = os.path.join(voc_dir, 'ImageSets', 'Segmentation',\n", " 'train.txt' if is_train else 'val.txt')\n", " mode = torchvision.io.image.ImageReadMode.RGB\n", " with open(txt_fname, 'r') as f:\n", " images = f.read().split()\n", " features, labels = [], []\n", " for i, fname in enumerate(images):\n", " features.append(torchvision.io.read_image(os.path.join(\n", " voc_dir, 'JPEGImages', f'{fname}.jpg')))\n", " labels.append(torchvision.io.read_image(os.path.join(\n", " voc_dir, 'SegmentationClass' ,f'{fname}.png'), mode))\n", " return features, labels\n", "\n", "train_features, train_labels = read_voc_images(voc_dir, True)" ] }, { "cell_type": "markdown", "id": "de24661e", "metadata": { "origin_pos": 8 }, "source": [ "Chúng tôi [** vẽ năm hình ảnh đầu vào đầu tiên và nhãn của chúng**]. Trong hình ảnh nhãn, màu trắng và đen đại diện cho đường viền và nền tương ứng, trong khi các màu khác tương ứng với các lớp khác nhau.\n" ] }, { "cell_type": "code", "execution_count": 4, "id": "3d4dab18", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:58.608826Z", "iopub.status.busy": "2022-04-26T12:32:58.608320Z", "iopub.status.idle": "2022-04-26T12:32:59.268476Z", "shell.execute_reply": "2022-04-26T12:32:59.267757Z" }, "origin_pos": 10, "tab": [ "pytorch" ] }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "n = 5\n", "imgs = train_features[0:n] + train_labels[0:n]\n", "imgs = [img.permute(1,2,0) for img in imgs]\n", "d2l.show_images(imgs, 2, n);" ] }, { "cell_type": "markdown", "id": "1951deba", "metadata": { "origin_pos": 11 }, "source": [ "Tiếp theo, chúng ta [** liệt kê các giá trị màu RGB và tên lớp **] cho tất cả các nhãn trong tập dữ liệu này.\n" ] }, { "cell_type": "code", "execution_count": 5, "id": "cd1e7c07", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.273093Z", "iopub.status.busy": "2022-04-26T12:32:59.272680Z", "iopub.status.idle": "2022-04-26T12:32:59.279473Z", "shell.execute_reply": "2022-04-26T12:32:59.278789Z" }, "origin_pos": 12, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "VOC_COLORMAP = [[0, 0, 0], [128, 0, 0], [0, 128, 0], [128, 128, 0],\n", " [0, 0, 128], [128, 0, 128], [0, 128, 128], [128, 128, 128],\n", " [64, 0, 0], [192, 0, 0], [64, 128, 0], [192, 128, 0],\n", " [64, 0, 128], [192, 0, 128], [64, 128, 128], [192, 128, 128],\n", " [0, 64, 0], [128, 64, 0], [0, 192, 0], [128, 192, 0],\n", " [0, 64, 128]]\n", "\n", "#@save\n", "VOC_CLASSES = ['background', 'aeroplane', 'bicycle', 'bird', 'boat',\n", " 'bottle', 'bus', 'car', 'cat', 'chair', 'cow',\n", " 'diningtable', 'dog', 'horse', 'motorbike', 'person',\n", " 'potted plant', 'sheep', 'sofa', 'train', 'tv/monitor']" ] }, { "cell_type": "markdown", "id": "1d029839", "metadata": { "origin_pos": 13 }, "source": [ "Với hai hằng số được định nghĩa ở trên, chúng ta có thể thuận tiện [** tìm chỉ mục lớp cho mỗi pixel trong một nhãy**]. Chúng tôi xác định hàm `voc_colormap2label` để xây dựng ánh xạ từ các giá trị màu RGB ở trên đến các chỉ số lớp và hàm `voc_label_indices` để ánh xạ bất kỳ giá trị RGB nào với chỉ số lớp của chúng trong bộ dữ liệu Pascal VOC2012 này.\n" ] }, { "cell_type": "code", "execution_count": 6, "id": "ababfe28", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.283070Z", "iopub.status.busy": "2022-04-26T12:32:59.282535Z", "iopub.status.idle": "2022-04-26T12:32:59.288397Z", "shell.execute_reply": "2022-04-26T12:32:59.287758Z" }, "origin_pos": 15, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "def voc_colormap2label():\n", " \"\"\"Build the mapping from RGB to class indices for VOC labels.\"\"\"\n", " colormap2label = torch.zeros(256 ** 3, dtype=torch.long)\n", " for i, colormap in enumerate(VOC_COLORMAP):\n", " colormap2label[\n", " (colormap[0] * 256 + colormap[1]) * 256 + colormap[2]] = i\n", " return colormap2label\n", "\n", "#@save\n", "def voc_label_indices(colormap, colormap2label):\n", " \"\"\"Map any RGB values in VOC labels to their class indices.\"\"\"\n", " colormap = colormap.permute(1, 2, 0).numpy().astype('int32')\n", " idx = ((colormap[:, :, 0] * 256 + colormap[:, :, 1]) * 256\n", " + colormap[:, :, 2])\n", " return colormap2label[idx]" ] }, { "cell_type": "markdown", "id": "55b2871f", "metadata": { "origin_pos": 16 }, "source": [ "[**Ví dụ**], trong ảnh ví dụ đầu tiên, chỉ mục lớp cho phần trước của máy bay là 1, trong khi chỉ mục nền là 0.\n" ] }, { "cell_type": "code", "execution_count": 7, "id": "796f50bd", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.292119Z", "iopub.status.busy": "2022-04-26T12:32:59.291612Z", "iopub.status.idle": "2022-04-26T12:32:59.500198Z", "shell.execute_reply": "2022-04-26T12:32:59.499358Z" }, "origin_pos": 17, "tab": [ "pytorch" ] }, "outputs": [ { "data": { "text/plain": [ "(tensor([[0, 0, 0, 0, 0, 0, 0, 0, 0, 1],\n", " [0, 0, 0, 0, 0, 0, 0, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 0, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 1, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 1, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 1, 1, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 1, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 1, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 0, 1, 1, 1, 1],\n", " [0, 0, 0, 0, 0, 0, 0, 0, 1, 1]]),\n", " 'aeroplane')" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "y = voc_label_indices(train_labels[0], voc_colormap2label())\n", "y[105:115, 130:140], VOC_CLASSES[1]" ] }, { "cell_type": "markdown", "id": "a4fd9485", "metadata": { "origin_pos": 18 }, "source": [ "### Xử lý sơ bộ dữ liệu\n", "\n", "Trong các thí nghiệm trước đó như trong :numref:`sec_alexnet`—:numref:`sec_googlenet`, hình ảnh được thay đổi lại để phù hợp với hình dạng đầu vào yêu cầu của mô hình. Tuy nhiên, trong phân đoạn ngữ nghĩa, làm như vậy đòi hỏi phải thay đổi lại các lớp pixel dự đoán trở lại hình dạng ban đầu của hình ảnh đầu vào. Việc tái cặn như vậy có thể không chính xác, đặc biệt là đối với các vùng được phân đoạn với các lớp khác nhau. Để tránh sự cố này, chúng tôi cắt hình ảnh thành một hình dạng *fixed* thay vì rescaling. Cụ thể, [** sử dụng cắt xén ngẫu nhiên từ nâng hình ảnh, chúng ta cắt cùng một vùng của hình ảnh đầu vào và nhãn**].\n" ] }, { "cell_type": "code", "execution_count": 8, "id": "89c5f626", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.503862Z", "iopub.status.busy": "2022-04-26T12:32:59.503328Z", "iopub.status.idle": "2022-04-26T12:32:59.508372Z", "shell.execute_reply": "2022-04-26T12:32:59.507516Z" }, "origin_pos": 20, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "def voc_rand_crop(feature, label, height, width):\n", " \"\"\"Randomly crop both feature and label images.\"\"\"\n", " rect = torchvision.transforms.RandomCrop.get_params(\n", " feature, (height, width))\n", " feature = torchvision.transforms.functional.crop(feature, *rect)\n", " label = torchvision.transforms.functional.crop(label, *rect)\n", " return feature, label" ] }, { "cell_type": "code", "execution_count": 9, "id": "13fffd23", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.511935Z", "iopub.status.busy": "2022-04-26T12:32:59.511238Z", "iopub.status.idle": "2022-04-26T12:32:59.993312Z", "shell.execute_reply": "2022-04-26T12:32:59.992578Z" }, "origin_pos": 22, "tab": [ "pytorch" ] }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "imgs = []\n", "for _ in range(n):\n", " imgs += voc_rand_crop(train_features[0], train_labels[0], 200, 300)\n", "\n", "imgs = [img.permute(1, 2, 0) for img in imgs]\n", "d2l.show_images(imgs[::2] + imgs[1::2], 2, n);" ] }, { "cell_type": "markdown", "id": "1eeed50e", "metadata": { "origin_pos": 23 }, "source": [ "### [**Custom Semantic Segmentation Dataset Class**]\n", "\n", "Chúng tôi xác định một tập dữ liệu phân đoạn ngữ nghĩa tùy chỉnh lớp `VOCSegDataset` bằng cách kế thừa lớp `Dataset` được cung cấp bởi các API cấp cao. Bằng cách thực hiện hàm `__getitem__`, chúng ta có thể tùy ý truy cập vào hình ảnh đầu vào được lập chỉ mục là `idx` trong tập dữ liệu và chỉ số lớp của mỗi pixel trong hình ảnh này. Vì một số hình ảnh trong tập dữ liệu có kích thước nhỏ hơn kích thước đầu ra của cắt ngẫu nhiên, các ví dụ này được lọc ra bởi một hàm `filter` tùy chỉnh. Ngoài ra, chúng tôi cũng xác định hàm `normalize_image` để chuẩn hóa các giá trị của ba kênh RGB của hình ảnh đầu vào.\n" ] }, { "cell_type": "code", "execution_count": 10, "id": "46ad80bd", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:32:59.997264Z", "iopub.status.busy": "2022-04-26T12:32:59.996742Z", "iopub.status.idle": "2022-04-26T12:33:00.005073Z", "shell.execute_reply": "2022-04-26T12:33:00.004412Z" }, "origin_pos": 25, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "class VOCSegDataset(torch.utils.data.Dataset):\n", " \"\"\"A customized dataset to load the VOC dataset.\"\"\"\n", "\n", " def __init__(self, is_train, crop_size, voc_dir):\n", " self.transform = torchvision.transforms.Normalize(\n", " mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n", " self.crop_size = crop_size\n", " features, labels = read_voc_images(voc_dir, is_train=is_train)\n", " self.features = [self.normalize_image(feature)\n", " for feature in self.filter(features)]\n", " self.labels = self.filter(labels)\n", " self.colormap2label = voc_colormap2label()\n", " print('read ' + str(len(self.features)) + ' examples')\n", "\n", " def normalize_image(self, img):\n", " return self.transform(img.float() / 255)\n", "\n", " def filter(self, imgs):\n", " return [img for img in imgs if (\n", " img.shape[1] >= self.crop_size[0] and\n", " img.shape[2] >= self.crop_size[1])]\n", "\n", " def __getitem__(self, idx):\n", " feature, label = voc_rand_crop(self.features[idx], self.labels[idx],\n", " *self.crop_size)\n", " return (feature, voc_label_indices(label, self.colormap2label))\n", "\n", " def __len__(self):\n", " return len(self.features)" ] }, { "cell_type": "markdown", "id": "3fbd0bdb", "metadata": { "origin_pos": 26 }, "source": [ "### [**Đọc dữ liệu**]\n", "\n", "Chúng tôi sử dụng lớp `VOCSegDatase`t tùy chỉnh để tạo các phiên bản của bộ đào tạo và bộ kiểm tra, tương ứng. Giả sử rằng chúng ta chỉ định rằng hình dạng đầu ra của hình ảnh được cắt ngẫu nhiên là $320\\times 480$. Dưới đây chúng ta có thể xem số ví dụ được giữ lại trong bộ đào tạo và bộ kiểm tra.\n" ] }, { "cell_type": "code", "execution_count": 11, "id": "f9a91ecd", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:33:00.008348Z", "iopub.status.busy": "2022-04-26T12:33:00.007750Z", "iopub.status.idle": "2022-04-26T12:33:12.923289Z", "shell.execute_reply": "2022-04-26T12:33:12.922341Z" }, "origin_pos": 27, "tab": [ "pytorch" ] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "read 1114 examples\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "read 1078 examples\n" ] } ], "source": [ "crop_size = (320, 480)\n", "voc_train = VOCSegDataset(True, crop_size, voc_dir)\n", "voc_test = VOCSegDataset(False, crop_size, voc_dir)" ] }, { "cell_type": "markdown", "id": "d2e1856f", "metadata": { "origin_pos": 28 }, "source": [ "Đặt kích thước lô thành 64, chúng tôi xác định bộ lặp dữ liệu cho bộ đào tạo. Hãy để chúng tôi in hình dạng của minibatch đầu tiên. Khác với phân loại hình ảnh hoặc phát hiện đối tượng, nhãn ở đây là hàng chục ba chiều.\n" ] }, { "cell_type": "code", "execution_count": 12, "id": "7089b316", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:33:12.926990Z", "iopub.status.busy": "2022-04-26T12:33:12.926744Z", "iopub.status.idle": "2022-04-26T12:33:14.557837Z", "shell.execute_reply": "2022-04-26T12:33:14.557011Z" }, "origin_pos": 30, "tab": [ "pytorch" ] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "torch.Size([64, 3, 320, 480])\n", "torch.Size([64, 320, 480])\n" ] } ], "source": [ "batch_size = 64\n", "train_iter = torch.utils.data.DataLoader(voc_train, batch_size, shuffle=True,\n", " drop_last=True,\n", " num_workers=d2l.get_dataloader_workers())\n", "for X, Y in train_iter:\n", " print(X.shape)\n", " print(Y.shape)\n", " break" ] }, { "cell_type": "markdown", "id": "fc39be27", "metadata": { "origin_pos": 31 }, "source": [ "### [** Putting All Things Together**]\n", "\n", "Cuối cùng, chúng tôi xác định hàm `load_data_voc` sau để tải xuống và đọc tập dữ liệu phân đoạn ngữ nghĩa Pascal VOC2012. Nó trả về bộ lặp dữ liệu cho cả tập dữ liệu đào tạo và kiểm tra.\n" ] }, { "cell_type": "code", "execution_count": 13, "id": "6a7e2be6", "metadata": { "execution": { "iopub.execute_input": "2022-04-26T12:33:14.562714Z", "iopub.status.busy": "2022-04-26T12:33:14.562434Z", "iopub.status.idle": "2022-04-26T12:33:14.568849Z", "shell.execute_reply": "2022-04-26T12:33:14.567961Z" }, "origin_pos": 33, "tab": [ "pytorch" ] }, "outputs": [], "source": [ "#@save\n", "def load_data_voc(batch_size, crop_size):\n", " \"\"\"Load the VOC semantic segmentation dataset.\"\"\"\n", " voc_dir = d2l.download_extract('voc2012', os.path.join(\n", " 'VOCdevkit', 'VOC2012'))\n", " num_workers = d2l.get_dataloader_workers()\n", " train_iter = torch.utils.data.DataLoader(\n", " VOCSegDataset(True, crop_size, voc_dir), batch_size,\n", " shuffle=True, drop_last=True, num_workers=num_workers)\n", " test_iter = torch.utils.data.DataLoader(\n", " VOCSegDataset(False, crop_size, voc_dir), batch_size,\n", " drop_last=True, num_workers=num_workers)\n", " return train_iter, test_iter" ] }, { "cell_type": "markdown", "id": "c7256461", "metadata": { "origin_pos": 34 }, "source": [ "## Tóm tắt\n", "\n", "* Phân đoạn ngữ nghĩa nhận ra và hiểu những gì trong một hình ảnh ở mức pixel bằng cách chia hình ảnh thành các vùng thuộc các lớp ngữ nghĩa khác nhau.\n", "* Trên của tập dữ liệu phân khúc ngữ nghĩa quan trọng nhất là Pascal VOC2012.\n", "* Trong phân đoạn ngữ nghĩa, vì hình ảnh đầu vào và nhãn tương ứng một-một trên pixel, hình ảnh đầu vào được cắt ngẫu nhiên thành một hình dạng cố định chứ không phải rescaled.\n", "\n", "## Bài tập\n", "\n", "1. Làm thế nào phân khúc ngữ nghĩa có thể được áp dụng trong các phương tiện tự trị và chẩn đoán hình ảnh y tế? Bạn có thể nghĩ về các ứng dụng khác?\n", "1. Nhớ lại các mô tả về tăng cường dữ liệu trong :numref:`sec_image_augmentation`. Phương pháp nâng hình ảnh nào được sử dụng trong phân loại hình ảnh sẽ không khả thi được áp dụng trong phân khúc ngữ nghĩa?\n" ] }, { "cell_type": "markdown", "id": "80b95d7e", "metadata": { "origin_pos": 36, "tab": [ "pytorch" ] }, "source": [ "[Discussions](https://discuss.d2l.ai/t/1480)\n" ] } ], "metadata": { "language_info": { "name": "python" } }, "nbformat": 4, "nbformat_minor": 5 }