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 Lagging behind the growth in model capacity, the available datasets are quickly becoming outdated in terms of size and density.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.
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
10 scene categories for LSUN Scene Classification Challange: Downloading Code
20 object categories: Link List. Images for each category are stored in LMDB format and the database is then zipped. After downloading and decompressing the zip files, please to refer to LSUN utility code to visualize and export the images. MD5 sum for each zip file is also provided so that you can verify your downloads.
In CVPR 2015 and 2016, a image classification challenge has been hosted in LSUN Challenge workshop to evaluate the progress of large-scale image understanding. More information can be found at the challenge webpage.