Deep Image: Scaling up Image Recognition

Abstract

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.

This paper has been withdrawn by the authors due to a mistake related to ImageNet server submissions

References in corpus (7)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• One weird trick for parallelizing convolutional neural networks
• Attention for Fine-Grained Categorization
• Deep convolutional filter banks for texture recognition and segmentation
• Generalized Max Pooling
• The Application of Two-level Attention Models in Deep Convolutional Neural Network for Fine-grained Image Classification

Cited by in corpus (24)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Improved Regularization of Convolutional Neural Networks with Cutout
• Deep Learning with Limited Numerical Precision
• Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networks
• A Joint Convolutional Neural Networks and Context Transfer for Street Scenes Labeling
• Further advantages of data augmentation on convolutional neural networks
• APAC: Augmented PAttern Classification with Neural Networks
• Analysis and Optimization of Convolutional Neural Network Architectures
• Hermite-Gaussian Mode Detection via Convolution Neural Networks
• Parle: parallelizing stochastic gradient descent
• Exploring the Design Space of Deep Convolutional Neural Networks at Large Scale
• Deep Action- and Context-Aware Sequence Learning for Activity Recognition and Anticipation
• Is Saki #delicious? The Food Perception Gap on Instagram and Its Relation to Health
• Low-cost and high-performance data augmentation for deep-learning-based skin lesion classification
• Can a CNN Recognize Catalan Diet?
• Learning from Learning Machines: Optimisation, Rules, and Social Norms
• Stanza: Layer Separation for Distributed Training in Deep Learning
• The Penalty Imposed by Ablated Data Augmentation
• Context Augmentation for Convolutional Neural Networks
• A Survey on Large-scale Machine Learning
• Explainability Guided Multi-Site COVID-19 CT Classification
• Efficient Multi-Person Pose Estimation with Provable Guarantees
• Cross-Domain Image Classification through Neural-Style Transfer Data Augmentation
• Hardware Synthesis of State-Space Equations; Application to FPGA Implementation of Shallow and Deep Neural Networks