Resolution Switchable Networks for Runtime Efficient Image Recognition
arXiv:2007.09558
Abstract
We propose a general method to train a single convolutional neural network which is capable of switching image resolutions at inference. Thus the running speed can be selected to meet various computational resource limits. Networks trained with the proposed method are named Resolution Switchable Networks (RS-Nets). The basic training framework shares network parameters for handling images which differ in resolution, yet keeps separate batch normalization layers. Though it is parameter-efficient in design, it leads to inconsistent accuracy variations at different resolutions, for which we provide a detailed analysis from the aspect of the train-test recognition discrepancy. A multi-resolution ensemble distillation is further designed, where a teacher is learnt on the fly as a weighted ensemble over resolutions. Thanks to the ensemble and knowledge distillation, RS-Nets enjoy accuracy improvements at a wide range of resolutions compared with individually trained models. Extensive experiments on the ImageNet dataset are provided, and we additionally consider quantization problems. Code and models are available at https://github.com/yikaiw/RS-Nets.
arXiv commentECCV 2020. Code and models: https://github.com/yikaiw/RS-Nets
References in corpus (8)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Distilling the Knowledge in a Neural Network
- MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
- Revisiting Batch Normalization For Practical Domain Adaptation
- Knowledge Distillation by On-the-Fly Native Ensemble
- K for the Price of 1: Parameter-efficient Multi-task and Transfer Learning
- Scale Calibrated Training: Improving Generalization of Deep Networks via Scale-Specific Normalization