精华内容
下载资源
问答
  • Dataset之LSUN:LSUN数据集的简介、安装、使用方法之详细攻略 目录 LSUN数据集的简介 1、Paper 2、简介 3、LSUN数据集上DCGAN的生成结果 LSUN数据集的安装 LSUN数据集的使用方法 LSUN数据集的简介...

    Dataset之LSUN:LSUN数据集的简介、安装、使用方法之详细攻略

     

     

    目录

    LSUN数据集的简介

    1、Paper

    2、简介

    3、LSUN数据集上DCGAN的生成结果

    LSUN数据集的安装

    LSUN数据集的使用方法


     

     

     

    LSUN数据集的简介

    1、Paper

    Fisher Yu, Ari Seff, Yinda Zhang, Shuran Song, Thomas Funkhouser and Jianxiong Xiao 
    LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
    arXiv:1506.03365 [cs.CV], 10 Jun 2015 

    2、简介

          LSun场景分类的10个场景类别。LSUN 是一个场景理解图像数据集,主要包含了卧室、固房、客厅、教室等场景图像。
          20对象类别:链接列表。每个类别的图像以LMDB格式存储,然后数据库被压缩。下载和解压缩ZIP文件后,请参考LSun实用代码来可视化和导出图像。还提供了每个zip文件的MD5和,以便您可以验证下载。

         While there has been remarkable progress in the performance of visual recognition algorithms, the state-of-the-art models tend to be exceptionally data-hungry. Large labeled training datasets, expensive and tedious to produce, are required to optimize millions of parameters in deep network models. Lagging behind the growth in model capacity, the available datasets are quickly becoming outdated in terms of size and density. To circumvent this bottleneck, we propose to amplify human effort through a partially automated labeling scheme, leveraging deep learning with humans in the loop. Starting from a large set of candidate images for each category, we iteratively sample a subset, ask people to label them, classify the others with a trained model, split the set into positives, negatives, and unlabeled based on the classification confidence, and then iterate with the unlabeled set. To assess the effectiveness of this cascading procedure and enable further progress in visual recognition research, we construct a new image dataset, LSUN. It contains around one million labeled images for each of 10 scene categories and 20 object categories. We experiment with training popular convolutional networks and find that they achieve substantial performance gains when trained on this dataset.

           虽然在视觉识别算法的性能上已经取得了显著的进步,但是最先进的模型往往特别需要数据。为了在深层网络模型中优化数百万个参数,需要大量标注的训练数据集,这些数据集的生产既昂贵又繁琐。滞后于模型容量的增长,可用的数据集在尺寸和密度方面很快变得过时。为了绕过这个瓶颈,我们建议通过部分自动化的标签方案,利用循环中的人的深层学习,来增强人的努力。从每个类别的一大组候选图像开始,我们迭代地采样一个子集,要求人们标记它们,用训练好的模型对其他类别进行分类,根据分类置信度将集合划分为正、负和未标记,然后用未标记的集合进行迭代。为了评估这种级联过程的有效性,并使视觉识别研究取得进一步进展,我们构建了一个新的图像数据集,LSUN。
          它包含10个场景类别和20个对象类别中的每一个的大约一百万个标记图像。我们对当前流行的卷积网络进行了实验,发现当在这个数据集上进行训练时,它们获得了显著的性能增益。

    官网地址:http://www.yf.io/p/lsun

    3、LSUN数据集上DCGAN的生成结果

     

     

    LSUN数据集的下载

            一个类别中的所有图像都存储在一个lmdb数据库文件中。每个条目的值是jpg二进制数据。我们调整所有的图像大小,使较小的尺寸是256和压缩的质量为75的jpeg图像。

     

    该数据集,暂时无法在线下载
    PS:如需该数据集,可向博主留言索取!

    Dataset之LSUN:LSUN数据集的下载使用教程

     

     

    LSUN数据集的使用方法

    基于LSUN数据集实现场景分类识别

     

     

     

     

     

     

    展开全文
  • 如果您发现LSUN数据集对您的研究有用,请考虑引用: @article{yu15lsun, Author = {Yu, Fisher and Zhang, Yinda and Song, Shuran and Seff, Ari and Xiao, Jianxiong}, Title = {LSUN: Construction of a ...
  • LSUN数据集(bedroom子集)

    2020-04-27 11:28:05
    LSUN数据集(bedroom),文件中包含百度网盘提取码。LSUN是一个比较有特色的大规模数据集,相对ImageNet而言,分类更丰富,不仅有物体分类,还有场景分类。
  • Dataset之LSUN:LSUN数据集的在线下载使用教程 LSUN数据集的在线下载代码 源文件代码:https://github.com/fyu/lsun 1、category_indices.txt bedroom 0 bridge 1 church_outdoor 2 classroom 3 ...

    Dataset之LSUN:LSUN数据集的在线下载使用教程

     

    目录

    LSUN数据集的在线下载代码

    1、category_indices.txt

    2、data.py

    3、download.py

    4、README.md


     

    LSUN数据集的在线下载代码

    curl方法下载

    curl http://dl.yf.io/lsun/scenes/bedroom_train_lmdb.zip -o data\\bedroom_train_lmdb.zip

     

    源文件代码https://github.com/fyu/lsun

    1、category_indices.txt

    bedroom 0
    bridge 1
    church_outdoor 2
    classroom 3
    conference_room 4
    dining_room 5
    kitchen 6
    living_room 7
    restaurant 8
    tower 9
    

     

    2、data.py

    from __future__ import print_function
    import argparse
    import cv2
    import lmdb
    import numpy
    import os
    from os.path import exists, join
    
    __author__ = 'Fisher Yu'
    __email__ = 'fy@cs.princeton.edu'
    __license__ = 'MIT'
    
    
    def view(db_path):
        print('Viewing', db_path)
        print('Press ESC to exist or SPACE to advance.')
        window_name = 'LSUN'
        cv2.namedWindow(window_name)
        env = lmdb.open(db_path, map_size=1099511627776,
                        max_readers=100, readonly=True)
        with env.begin(write=False) as txn:
            cursor = txn.cursor()
            for key, val in cursor:
                print('Current key:', key)
                img = cv2.imdecode(
                    numpy.fromstring(val, dtype=numpy.uint8), 1)
                cv2.imshow(window_name, img)
                c = cv2.waitKey()
                if c == 27:
                    break
    
    
    def export_images(db_path, out_dir, flat=False, limit=-1):
        print('Exporting', db_path, 'to', out_dir)
        env = lmdb.open(db_path, map_size=1099511627776,
                        max_readers=100, readonly=True)
        count = 0
        with env.begin(write=False) as txn:
            cursor = txn.cursor()
            for key, val in cursor:
                if not flat:
                    image_out_dir = join(out_dir, '/'.join(key[:6]))
                else:
                    image_out_dir = out_dir
                if not exists(image_out_dir):
                    os.makedirs(image_out_dir)
                image_out_path = join(image_out_dir, key + '.webp')
                with open(image_out_path, 'w') as fp:
                    fp.write(val)
                count += 1
                if count == limit:
                    break
                if count % 1000 == 0:
                    print('Finished', count, 'images')
    
    
    def main():
        parser = argparse.ArgumentParser()
        parser.add_argument('command', nargs='?', type=str,
                            choices=['view', 'export'],
                            help='view: view the images in the lmdb database '
                                 'interactively.\n'
                                 'export: Export the images in the lmdb databases '
                                 'to a folder. The images are grouped in subfolders'
                                 ' determinted by the prefiex of image key.')
        parser.add_argument('lmdb_path', nargs='+', type=str,
                            help='The path to the lmdb database folder. '
                                 'Support multiple database paths.')
        parser.add_argument('--out_dir', type=str, default='')
        parser.add_argument('--flat', action='store_true',
                            help='If enabled, the images are imported into output '
                                 'directory directly instead of hierarchical '
                                 'directories.')
        args = parser.parse_args()
    
        command = args.command
        lmdb_paths = args.lmdb_path
    
        for lmdb_path in lmdb_paths:
            if command == 'view':
                view(lmdb_path)
            elif command == 'export':
                export_images(lmdb_path, args.out_dir, args.flat)
    
    
    if __name__ == '__main__':
        main()
    

     

    3、download.py

    # -*- coding: utf-8 -*-
    
    from __future__ import print_function, division
    import argparse
    from os.path import join
    
    import subprocess
    from urllib.request import Request, urlopen
    
    __author__ = 'Fisher Yu'
    __email__ = 'fy@cs.princeton.edu'
    __license__ = 'MIT'
    
    
    def list_categories():
        url = 'http://dl.yf.io/lsun/categories.txt'
        with urlopen(Request(url)) as response:
            return response.read().decode().strip().split('\n')
    
    
    def download(out_dir, category, set_name):
        url = 'http://dl.yf.io/lsun/scenes/{category}_' \
              '{set_name}_lmdb.zip'.format(**locals())
        if set_name == 'test':
            out_name = 'test_lmdb.zip'
            url = 'http://dl.yf.io/lsun/scenes/{set_name}_lmdb.zip'
        else:
            out_name = '{category}_{set_name}_lmdb.zip'.format(**locals())
        out_path = join(out_dir, out_name)
        print('out_path:',out_path)
        cmd = ['curl', url, '-o', out_path]
        print(cmd)
        print('Downloading', category, set_name, 'set')
        subprocess.call(cmd)
    
    
    def main():
        parser = argparse.ArgumentParser()
        parser.add_argument('-o', '--out_dir', default='')
        parser.add_argument('-c', '--category', default=None)
        args = parser.parse_args()
    
        categories = list_categories()
        print('categories',categories)
        if args.category is None:
            print('Downloading', len(categories), 'categories')
            for category in categories:
                download(args.out_dir, category, 'train')
                download(args.out_dir, category, 'val')
            download(args.out_dir, '', 'test')
        else:
            if args.category == 'test':
                download(args.out_dir, '', 'test')
            elif args.category not in categories:
                print('Error:', args.category, "doesn't exist in", 'LSUN release')
            else:
                download(args.out_dir, args.category, 'train')
                download(args.out_dir, args.category, 'val')
    
    
    if __name__ == '__main__':
        main()
    

     

    4、README.md

    # LSUN
    
    Please check [LSUN webpage](http://www.yf.io/p/lsun) for more information about the dataset.
    
    ## Data Release
    
    All the images in one category are stored in one lmdb database
    file. The value
     of each entry is the jpg binary data. We resize all the images so
     that the
      smaller dimension is 256 and compress the images in jpeg with
      quality 75.
      
    ### Citing LSUN
    
    If you find LSUN dataset useful in your research, please consider citing:
    
        @article{yu15lsun,
            Author = {Yu, Fisher and Zhang, Yinda and Song, Shuran and Seff, Ari and Xiao, Jianxiong},
            Title = {LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop},
            Journal = {arXiv preprint arXiv:1506.03365},
            Year = {2015}
        }
    
    ### Download data
    Please make sure you have cURL installed
    ```bash
    # Download the whole latest data set
    python3 download.py
    # Download the whole latest data set to <data_dir>
    python3 download.py -o <data_dir>
    # Download data for bedroom
    python3 download.py -c bedroom
    # Download testing set
    python3 download.py -c test
    ```
    
    ## Demo code
    
    ### Dependency
    
    Install Python
    
    Install Python dependency: numpy, lmdb, opencv
    
    ### Usage:
    
    View the lmdb content
    
    ```bash
    python3 data.py view <image db path>
    ```
    
    Export the images to a folder
    
    ```bash
    python3 data.py export <image db path> --out_dir <output directory>
    ```
    
    ### Example:
    
    Export all the images in valuation sets in the current folder to a
    "data"
    subfolder.
    
    ```bash
    python3 data.py export *_val_lmdb --out_dir data
    ```
    
    ## Submission
    
    We expect one category prediction for each image in the testing
    set. The name of each image is the key value in the LMDB
    database. Each category has an index as listed in
    [index list](https://github.com/fyu/lsun_toolkit/blob/master/category_indices.txt). The
    submitted results on the testing set will be stored in a text file
    with one line per image. In each line, there are two fields separated
    by a whitespace. The first is the image key and the second is the
    predicted category index. For example:
    
    ```
    0001c44e5f5175a7e6358d207660f971d90abaf4 0
    000319b73404935eec40ac49d1865ce197b3a553 1
    00038e8b13a97577ada8a884702d607220ce6d15 2
    00039ba1bf659c30e50b757280efd5eba6fc2fe1 3
    ...
    ```
    
    The score for the submission is the percentage of correctly predicted
    labels. In our evaluation, we will double check our ground truth
    labels for the testing images and we may remove some images with
    controversial labels in the final evaluation.
    

     

     

     

     

     

     

     

     

     

    展开全文
  • LSUN数据集 百度云【持续更新】 文章目录LSUN数据集 百度云【持续更新】LSUN类别bedroom LSUN 提示:本数据集于2020年9月从LSUN官网下载 类别 bedroom 链接:https://pan.baidu.com/s/1uLnXFS6cHuvNL5Ry2XpR6Q ...

    LSUN数据集 百度云【持续更新】


    提示:本数据集于2020年9月从LSUN官网下载


    类别

    bedroom【42.78G】

    链接:https://pan.baidu.com/s/1uLnXFS6cHuvNL5Ry2XpR6Q
    提取码:oq6j

    bridge【15.35G】

    链接:https://pan.baidu.com/s/1utGNfvfqWJwDFDNUazSMDg
    提取码:vukp

    church_outdoor【2.29G】

    链接:https://pan.baidu.com/s/1fNA5g3jxkzYA0-tU-uaLYQ
    提取码:epsw

    kitchen【33.34G】

    链接:https://pan.baidu.com/s/1r1W4PJte7cuqNAinwfNFYg
    提取码:1hv3

    dining_room【10.8G】

    链接:https://pan.baidu.com/s/1VNTj3tmiXp3t0gAASZWtJQ
    提取码:d7o2

    conference_room【3.78G】

    链接:https://pan.baidu.com/s/10XyTp-5Y_t_SZcQP8h_-FQ
    提取码:deiu

    tower【11.9G】

    链接:https://pan.baidu.com/s/1ACktFMO6PmuZMi60OPqM4g
    提取码:y82h

    classroom【3.06G】

    链接:https://pan.baidu.com/s/1VP3wzY9isQEP8vEEXhmgNQ
    提取码:2m75

    restaurant【12.57G】

    链接:https://pan.baidu.com/s/17HH8X_-3PE41UiHP2P29bw
    提取码:w0t3

    living_room【21.24G】

    链接:https://pan.baidu.com/s/1i0GKR0S1-D0i-6aPMUbNSQ
    提取码:1kpt

    展开全文
  • 最近复现SAGAN用到了lsun数据集 1.下载地址http://dl.yf.io/lsun/scenes/ 我下载了church_outdoor_train_lmdb.zip 中间还安装了lmdb库 2.转换 一般下载的data.py(转换代码)有bug,这段代码已经进行了修复,...

    最近复现SAGAN用到了lsun数据集

    1.下载地址  http://dl.yf.io/lsun/scenes/   我下载了church_outdoor_train_lmdb.zip

    中间还安装了lmdb库

    2.转换

    一般下载的data.py(转换代码)有bug,这段代码已经进行了修复,编译脚本并运行即可

     python data.py export ./church_outdoor_train_lmdb --out_dir ./lsun --flat

    # -*- coding: utf-8 -*-
    """
    Created on Mon Sep 28 10:28:13 2020
    
    @author: ZM
    """
    
    from __future__ import print_function
    import argparse
    import cv2
    import lmdb
    import numpy
    import os
    from os.path import exists, join
    
    __author__ = 'Fisher Yu'
    __email__ = 'fy@cs.princeton.edu'
    __license__ = 'MIT'
    
    
    def view(db_path):
        print('Viewing', db_path)
        print('Press ESC to exist or SPACE to advance.')
        window_name = 'LSUN'
        cv2.namedWindow(window_name)
        env = lmdb.open(db_path, map_size=1099511627776,
                        max_readers=100, readonly=True)
        with env.begin(write=False) as txn:
            cursor = txn.cursor()
            for key, val in cursor:
                print('Current key:', key)
                img = cv2.imdecode(
                    numpy.fromstring(val, dtype=numpy.uint8), 1)
                cv2.imshow(window_name, img)
                c = cv2.waitKey()
                if c == 27:
                    break
    
    
    def export_images(db_path, out_dir, flat=False, limit=-1):
        print('Exporting', db_path, 'to', out_dir)
        env = lmdb.open(db_path, map_size=1099511627776,
                        max_readers=100, readonly=True)
        count = 0
        with env.begin(write=False) as txn:
            cursor = txn.cursor()
            for key, val in cursor:
                if not flat:
                    image_out_dir = join(out_dir, '/'.join(key[:6].decode()))
                else:
                    image_out_dir = out_dir
                if not exists(image_out_dir):
                    os.makedirs(image_out_dir)
                image_out_path = join(image_out_dir, key.decode() + '.jpg')
                with open(image_out_path, 'wb') as fp:
                    fp.write(val)
                count += 1
                if count == limit:
                    break
                if count % 1000 == 0:
                    print('Finished', count, 'images')
    
    
    def main():
        parser = argparse.ArgumentParser()
        parser.add_argument('command', nargs='?', type=str,
                            choices=['view', 'export'],
                            help='view: view the images in the lmdb database '
                                 'interactively.\n'
                                 'export: Export the images in the lmdb databases '
                                 'to a folder. The images are grouped in subfolders'
                                 ' determinted by the prefiex of image key.')
        parser.add_argument('lmdb_path', nargs='+', type=str,
                            help='The path to the lmdb database folder. '
                                 'Support multiple database paths.')
        parser.add_argument('--out_dir', type=str, default='')
        parser.add_argument('--flat', action='store_true',
                            help='If enabled, the images are imported into output '
                                 'directory directly instead of hierarchical '
                                 'directories.')
        args = parser.parse_args()
    
        command = args.command
        lmdb_paths = args.lmdb_path
    
        for lmdb_path in lmdb_paths:
            if command == 'view':
                view(lmdb_path)
            elif command == 'export':
                export_images(lmdb_path, args.out_dir, args.flat)
    
    
    if __name__ == '__main__':
        main()
    

     3.根据具体代码使用lmdb格式或者jpg格式

     转换成功后可以进行重命名,全选右键重命名,对第一个命名为1,然后双击bat文件

    @Echo Off&SetLocal ENABLEDELAYEDEXPANSION
    FOR %%a in (*) do (
    set "name=%%a"
    set "name=!name: (=!"
    set "name=!name:)=!"
    ren "%%a" "!name!"
    )
    exit

     

    展开全文
  • LSUN数据集 类似ImageNet的大规模数据集,最早出自这篇论文: 《LSUN: Construction of a Large-Scale Image Dataset using Deep Learning with Humans in the Loop》 这个数据集借助深度学习自动标注图片,相比...
  • 类似ImageNet的大规模数据集,相对ImageNet,LSUN分类更丰富,不仅有物品分类,也要场景分类,下载文件内附百度云盘提取码。
  • 类似ImageNet的大规模数据集,相对ImageNet,LSUN分类更丰富,不仅有物品分类,也要场景分类,下载文件内附百度云盘提取码。
  • DCGAN数据集:mnist、CelebA、lsun

    千次阅读 2018-07-20 16:58:19
    carpedm20/DCGAN-tensorflow : ... mnist数据集: http://yann.lecun.com/exdb/mnist/ CelebA数据集: https://blog.csdn.net/Cloudox_/article/details/78432517?locat...
  • AI数据集

    千次阅读 2017-02-22 14:59:44
    数据集计算机视觉 MNIST CIFAR  ImageNet LSUN PASCAL SVHN MSCOCO Genome Faces 自然语言处理 语言模型  语义相似性 文本分类数据集 问答 阅读理解 情感分析 IR Maluuba   语音 推荐和排序系统 网络和图表  ...
  • 图片数据集+使用数据+数据预处理

    千次阅读 2018-11-13 12:59:03
    1、The CIFAR-10 dataset 10类,一共含有60000张32*32的彩色图片,每类大概6000张,测试集大概1000张,5000张训练集 ...2、imageNet数据集 网址:http://image-net.org/ 3、ImageFolder   4、LSUN Classif...
  • 人工智能公开数据集汇总

    千次阅读 2018-07-12 17:13:34
    经典数据集: MNIST | CIFAR | PASCAL VOC | MS COCO | LSUN | SVHN   人类相关数据: 1)人脸特征点数据 IBUG(intelligent bahaviour understanding group) 2)人体姿态数据 MPII Human Pose Dataset  ...
  • 深度学习视觉常用数据集 ## 经典数据集 ImageNet: http://image-net.org/ LFW人脸数据库: http://vis-www.cs.umass.edu/lfw/lfw.tgz LSUN:场景理解与许多辅助任务(房间布局估计,显着性预测等) ...
  • pytorch为我们封装好了很多经典的数据集在torchvision.datasets包里, torchvision.datasets这个包中包含MNIST、FakeData、COCO、LSUN、ImageFolder、DatasetFolder、ImageNet、CIFAR等一些常用的数据集,并且提供了...
  • Computer Vision MNIST CIFAR 10 & CIFAR 100 ImageNet LSUN PASCAL VOC SVHN MS COCO Visual Genome Labeled Faces in the Wild
  • Torchvision 数据集 torchvision.datasets包含数据集: MNIST COCO(用于图像标注和目标检测)(Captioning and Detection) LSUN Classification ImageFolder Imagenet-12 CIFAR10 and CIFAR100 STL10 SVHN ...
  • torchvision.datasets中包含了以下数据集 MNIST COCO(用于图像标注和目标检测)(Captioning and Detection) LSUN Classification ImageFolder Imagenet-12 CIFAR10 and CIFAR100 STL10 详细介绍(以mnist手写数字...
  • Unsupervised representation learning with Deep Convolutional GAN 无监督深度卷积对抗生成网络–学习笔记(DCGAN:在MNIST数据集上使用Keras实现) ...文章的实验在MNIST数据集LSUN(Large-scale Scene Unde...
  • belme:数据集中包含大量有标注的图像数据。 ImageNet: 是一个用于视觉对象识别软件研究的大型可视化数据库。超过1400万的图像URL被ImageNet手动注释。根据 WordNet 层次结构来组织,其中层次结构的每个节点都由...
  • 成功解决ForkingPickler(file, protocol).dump(obj) TypeError:...本人是在调试lsun数据集&&神经网络代码时出现,问题显示如下: 因为windows操作系统的原因,在Windows中,多进程multiprocessing使用的是序
  • 数据加载和预处理

    2021-02-07 00:12:27
    文章目录DatasetDataloadertorchvision包torchvision.modelstorchvision.... 并且torchvision已经预先实现了常用图像数据集,包括前面使用过的CIFAR-10,ImageNet、COCO、MNIST、LSUN数据集,可通过torchvisio
  • PyTorch 通过 torch.utils....并且 torchvision 已经预先实现了常用图像数据集,包括前面使用过的 CIFAR-10, ImageNet, COCO, MNIST, LSUN数据集,可通过 torchvision.datasets 方便的调用。 Dataset Datase...
  • PyTorch 基础 :数据的加载和预处理 ... 并且torchvision已经预先实现了常用图像数据集,包括前面使用过的CIFAR-10,ImageNet、COCO、MNIST、LSUN数据集,可通过torchvision.datasets方便的调用 import torch...
  • 并且torchvision已经预先实现了常用图像数据集,包括前面使用过的CIFAR-10,ImageNet、COCO、MNIST、LSUN数据集,可通过torchvision.datasets方便的调用 # 首先要引入相关的包 import torch #打印一下版本 ...
  • 基于PyTorch的Vgg16训练数据

    千次阅读 2020-05-04 18:07:06
    torchvision已经预先实现了常用的Dataset,包括前面使用过的CIFAR-10,以及ImageNet、COCO、MNIST、LSUN数据集,可通过诸如torchvision.datasets.CIFAR10来调用。在这里介绍一个会经常使用到的Dataset——...
  • 【PyTorch】数据加载

    2020-06-05 12:19:31
    torchvision中实现了一些常用的数据集,可以通过torchvision.datasets直接调用。如:MNIST,COCO,Captions,Detection,LSUN,ImageFolder,Imagenet-12,CIFAR,STL10,SVHN,PhotoTour。 torchvi.
  • 1、torchvision已经余弦实现了常用的Dataset,包括CIFAR-10,以及ImageNet、COCO、MNIST、LSUN数据集,可通过诸如torchvision.datasets.CIFAR10来调用。 2、ImageFolder假设所有的文件按文件夹保存,每个文件下...

空空如也

空空如也

1 2 3 4
收藏数 61
精华内容 24
关键字:

lsun数据集