Skip to content
An Unoffical Implementation of PeleeNet by TensorFlow, Keras
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
images
summary
.gitignore
LICENSE
README.md
cifar10.py
pelee_net.py

README.md

PeleeNet-Keras

An Unoffical Implementation of PeleeNet by TensorFlow, Keras.
Implemented training with CIFAR-10.

Original Paper

R. J. Wang, X. Li, C. X. Ling. Pelee: A Real-Time Object Detection System on Mobile Devices. NIPS. 2018. https://arxiv.org/abs/1804.06882

The original implementation by Caffe. https://github.com/Robert-JunWang/Pelee

Get started

from pelee_net import PeleeNet
model = PeleeNet(input_shape=(224,224,3), use_stem_block=True, n_classes=1000)

Parameters

  • input_shape : Resolution of input images. 224x224x3 by default(same as the original).
  • use_stem_block : Whether to use Stem Block. If True it's same as the original, if False input is connected directly to the first Dense Layer.
  • n_classes : Number of classes in prediction

Results on CIFAR-10

Augmentation Stem Block No weight Decay 5e-4 Weight Decay
No No 0.7633 0.9247
No Yes 0.8280 0.9218
Yes No 0.8881 0.9446
Yes Yes 0.8996 0.9410
  • Enable stem block cases : Input=(224, 224, 3), Upsampling x7
  • Disable stem block cases : Input=(32, 32, 3), No upsampling

Data augmentation is the standard data augmentation(4 pixels shift and horizontal flip).

No weight decay

5e-4 weight decay

There was no discussion int the paper on weight decay. But I noticed that weight decay is important in increasing the accuracy of CIFAR-10, so I added it.

Details (Japanese)

DenseNetの軽量版、PeleeNetをKerasで実装した
https://qiita.com/koshian2/items/187e240f478504079e7a

You can’t perform that action at this time.