{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "b74163c1-4b23-4c40-9afd-f963e4a1377d", "metadata": {}, "outputs": [], "source": [ "from glob import glob\n", "from datasets import Dataset, Features, Image, Value\n", "from PIL import Image as PILImage\n", "from torchvision.transforms import CenterCrop, Resize\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "import torch" ] }, { "cell_type": "code", "execution_count": 2, "id": "912bc52b-2c88-421a-9d3d-d85c5a0f2a47", "metadata": {}, "outputs": [], "source": [ "LSDIR_paths = Dataset.from_dict({\"path\":glob(\"LSDIR/*/*.png\")})" ] }, { "cell_type": "code", "execution_count": 3, "id": "e4ffa17d-3a4e-4c84-bf1b-905001deff29", "metadata": {}, "outputs": [], "source": [ "lanczos = PILImage.Resampling.LANCZOS\n", "def load_image(sample):\n", " image_path = sample['path']\n", " image = PILImage.open(image_path)\n", " w = image.width\n", " h = image.height\n", " aspect = w/h\n", " if aspect <= 9/16:\n", " if w<756 or h<1344:\n", " image = CenterCrop((960,540))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((1344,756))(Resize(756,lanczos)(image))\n", " elif aspect <= 2/3:\n", " if w<816 or h<1224:\n", " image = CenterCrop((810,540))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((1224,816))(Resize(816,lanczos)(image))\n", " elif aspect >= 16/9:\n", " if w<1344 or h<756:\n", " image = CenterCrop((540,960))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((756,1344))(Resize(756,lanczos)(image))\n", " elif aspect >= 8/5:\n", " if w<1248 or h<780:\n", " image = CenterCrop((540,864))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((780,1248))(Resize(780,lanczos)(image))\n", " elif aspect >= 4/3:\n", " if w<1152 or h<864:\n", " image = CenterCrop((540,720))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((864,1152))(Resize(864,lanczos)(image))\n", " else:\n", " if w<1008 or h<1008:\n", " image = CenterCrop((540,540))(Resize(540,lanczos)(image))\n", " else:\n", " image = CenterCrop((1008,1008))(Resize(1008,lanczos)(image))\n", " return {\n", " \"image\": image,\n", " \"w\":image.width,\n", " \"h\":image.height,\n", " \"mode\":image.mode,\n", " \"aspect\":image.width/image.height,\n", " \"n_pixels\":image.width*image.height}" ] }, { "cell_type": "code", "execution_count": 4, "id": "115873bf-a6ba-4a9d-a4c1-c2507b558e39", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "c0f5cfc20f7c4c4e88857d1bbefa1c84", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Map (num_proc=24): 0%| | 0/84991 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.hist(LSDIR['aspect'],bins=12, range=(0.5,2));" ] }, { "cell_type": "code", "execution_count": 8, "id": "0ee68902-8cbf-404f-9141-85230b8ef0e6", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.hist(LSDIR['h'],bins=12,alpha=0.7);\n", "plt.hist(LSDIR['w'],bins=12,alpha=0.7);" ] }, { "cell_type": "code", "execution_count": null, "id": "90f2874d-3f0d-4ad4-bcc1-3766fb87815a", "metadata": {}, "outputs": [], "source": [ "LSDIR.push_to_hub(\"danjacobellis/LSDIR\",split=\"train\") " ] }, { "cell_type": "code", "execution_count": 2, "id": "b442f513-e137-4800-a782-2256db2c784f", "metadata": {}, "outputs": [], "source": [ "LSDIR_paths = Dataset.from_dict({\"path\":glob(\"LSDIR_val/val1/HR/val/*.png\")})" ] }, { "cell_type": "code", "execution_count": 3, "id": "249ad020-35be-40c6-b6f9-f601b48eebae", "metadata": {}, "outputs": [], "source": [ "def load_image_val(sample):\n", " image_path = sample['path']\n", " image = PILImage.open(image_path)\n", " return {\n", " \"image\": image,\n", " \"w\":image.width,\n", " \"h\":image.height,\n", " \"mode\":image.mode,\n", " \"aspect\":image.width/image.height,\n", " \"n_pixels\":image.width*image.height}" ] }, { "cell_type": "code", "execution_count": 4, "id": "4ad4f210-1290-40dd-850b-de0ed8f8c05f", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "6d89cdec3c664a6c84e100e4828d6959", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Map (num_proc=24): 0%| | 0/250 [00:00