We present a state-of-the-art image recognition system, Deep Image, developed using end-to-end deep learning. The key components are a custom-built supercomputer dedicated to deep learning, a highly optimized parallel algorithm using new strategies for data partitioning and communication, larger deep neural network models, novel data augmentation approaches, and usage of multi-scale high-resolution images. Our method achieves excellent results on multiple challenging computer vision benchmarks.
@article{arxiv.1501.02876,
title = {Deep Image: Scaling up Image Recognition},
author = {Ren Wu and Shengen Yan and Yi Shan and Qingqing Dang and Gang Sun},
journal= {arXiv preprint arXiv:1501.02876},
year = {2015}
}
Comments
This paper has been withdrawn by the authors due to a mistake related to ImageNet server submissions