MSDNet

by gaohuang

gaohuang /MSDNet

Multi-Scale Dense Networks for Resource Efficient Image Classification (ICLR 2018 Oral)

429 Stars 93 Forks Last release: Not found MIT License 21 Commits 0 Releases

Available items

No Items, yet!

The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:

MSDNet

This repository provides the code for the paper Multi-Scale Dense Networks for Resource Efficient Image Classification.

Update on April 3, 2019 -- PyTorch implementation released!

A PyTorch implementation of MSDNet can be found from here.

Introduction

This paper studies convolutional networks that require limited computational resources at test time. We develop a new network architecture that performs on par with state-of-the-art convolutional networks, whilst facilitating prediction in two settings: (1) an anytime-prediction setting in which the network's prediction for one example is progressively updated, facilitating the output of a prediction at any time; and (2) a batch computational budget setting in which a fixed amount of computation is available to classify a set of examples that can be spent unevenly across 'easier' and 'harder' examples.

Figure 1: MSDNet layout (2D).

Figure 2: MSDNet layout (3D).

Results

(a) anytime-prediction setting

Figure 3: Anytime prediction on ImageNet.

(b) batch computational budget setting

Figure 4: Prediction under batch computational budget on ImageNet.

Figure 5: Random example images from the ImageNet classes

Red wine
and
Volcano
. Top row: images exited from the first classification layer of an MSDNet with correct prediction; Bottom row: images failed to be correctly classified at the first classifier but were correctly predicted and exited at the last layer.

Usage

Our code is written under the framework of Torch ResNet (https://github.com/facebook/fb.resnet.torch). The training scripts come with several options, which can be listed with the

--help
flag.
th main.lua --help

Configuration

In all the experiments, we use a validation set for model selection. We hold out

5000
training images on CIFAR, and
50000
images on ImageNet as the validation set.

Training recipe

Train an MSDNet with 10 classifiers attached to every other layer for anytime prediction:

bash
th main.lua -netType msdnet -dataset cifar10 -batchSize 64 -nEpochs 300 -nBlocks 10 -stepmode even -step 2 -base 4

Train an MSDNet with 7 classifiers with the span linearly increases for efficient batch computation:

bash
th main.lua -netType msdnet -dataset cifar10 -batchSize 64 -nEpochs 300 -nBlocks 7 -stepmode lin_grow -step 1 -base 1

Pre-trained ImageNet Models

  1. Download model checkpoints and the validation set indeces.

  2. Testing script:

    th main.lua -dataset imagenet -testOnly true -resume  -data  -gen 

FAQ

  1. How to calculate the FLOPs (or mul-add op) of a model?

We strongly recommend doing it automatically. Please refer to the op-counter project (LuaTorch), or the script in ConDenseNet (PyTorch). The basic idea of these op counters is to add a hook before the forward pass of a model.

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.