ImageNet

ImageNet Tutorial

This is a step by step example of AlexNet for ImageNet Challenge.

Introduction

This tutorial will guide you train your own super vision model. The default configure will take more than 3GB GPU RAM, so make batch size smaller or larger according to your GPU RAM size.

• Normally, smaller batch_size means more noise in gradient, and maybe a smaller learning rate is needed.
• If you want to still use a large batch with not enough RAM, You can set update_period=2 and batch_size=128, this means the parameter update is done every 2 batches, which is equivalent to batch_size=256

0.Before you start

Make sure you have downloaded the ImageNet training data. You don't need to resize the images by yourself, currently im2rec could resize it automatically. You could check the promoting message of im2rec for details.

1.Make the image list

After you get the data, you need to make a image list file first. The format is

integer_image_index \t label_index \t path_to_image

In general, the program will take a list of names of all image, shuffle them, then separate them into training files name list and testing file name list. Write down the list in the format.

A sample file is provided here

895099  464     n04467665_17283.JPEG
10025081        412     ILSVRC2010_val_00025082.JPEG
74181   789     n01915811_2739.JPEG
10035553        859     ILSVRC2010_val_00035554.JPEG
10048727        929     ILSVRC2010_val_00048728.JPEG
94028   924     n01980166_4956.JPEG
1080682 650     n11807979_571.JPEG
972457  633     n07723039_1627.JPEG
7534    11      n01630670_4486.JPEG
1191261 249     n12407079_5106.JPEG

2.Make the binary file

Although you can use image iterator now. the disk random seek will make the training process extremely slow. So you'd better generate binary file for training and use imgrec iterator .

To generate binary image, you need to use im2rec in the tool folder. The im2rec will take the path of image list file you generated just now, root path of the images and the output file path as input. These processes usually take several hours, so be patient. :)

A sample command:

./bin/im2rec image.lst image_root_dir output.bin resize=256

More details can be found by running ./bin/im2rec.

3.Set correct configuration file

Change the iterator path in the ImageNet.conf or kaiming.conf to point to your image list file and image binary file correctly, then just run as MNIST example. After about 20 round, you can see some reasonable result. We strongly recommend to use kaiming.conf, since it could provide much better results than Alexnet, while keeping the time cost unchanged. By calling

cxxnet ./kaiming.conf 2>eval.log

You can save the evaluation result into the file eval.log

=

Acknowledgment

• Reference: Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "ImageNet Classification with Deep Convolutional Neural Networks." NIPS. Vol. 1. No. 2. 2012.
• Reference: He, Kaiming, and Jian Sun. "Convolutional Neural Networks at Constrained Time Cost." CVPR2015
• The network parameter scheduling is adapted from configuration provided by Caffe