Skip to content

nickbiso/Keras-Caltech-256

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Caltech-256

Caltech-256 is a challenging set of 257 (including the last category of clutter) object categories containing a total of only 30607 images. Furthermore this dataset is imbalanced as seen in the plot below. In this exercise I utilized different Neural Network architectures and compare their performance. This project took me exactly 1 month because of the scale of the problem and the training and tweaking multiple CNN models that most took overnight to train.

Sample Images

alt text

Category Distribution

alt text

Results

Model Accuracy(Test-set)
Fully Connected 14%
VGG 44%
Inception 45%
Resnet 48%
VGG (Transfer-Learning) 57%
Inception (Transfer-Learning) 63%
Inception Resnet (Transfer-Learning) 71%
Xception (Transfer-Learning) 66%

Conclusion

As seen in the table above using a Convolutional Neural Network is a big leap in accuracy from Fully Connected Neural Networks and Transfer learning of is a significantly better than training a CNN from scratch. Furthermore, Transfer Learning is the fastest to train because you are only training a fraction of the network and in addition, it requires the least data.

About

257-way Image Classification using Fully Connected Neural Network, Convolutional Neural Network built from scratch and Transfer Learning

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published