Skip to content

a120092009/CNN_DOG

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

CNN_DOG

Author Tianyi Zhao

This work is inspired by Kaggle Playground Prediction Competition: Dog Breed Identification (https://www.kaggle.com/c/dog-breed-identification)

Database

Contents of Stanford Dogs dataset:

Number of categories: 120

Number of images: 20,580

Annotations: Class labels, Bounding boxes

Training set: Contains 12,000 images, 120 categories, 100 images for each category.

Test Set: Contains 8,580 images, 120 categories.

Implementation

Use VGG19 which is pre-trained on ImageNet and transfer this model for our dog breed classification problem.

vgg_trainable.py: Construction of VGG19 network

Demo.py: Training VGG19

Test_Visualization.ipynb: Get visualization of our model.

Test.ipynb: Get result visualization on test set.

Experiment Setting:

batch size = 64

Initial learning rate 0.001, decay step = 100, decay rate = 0.9

Activation function: ReLu

Dropout: 0.5

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published