Skip to content

Cifar-10 object recognition challenge with simple CNN (89% accuracy after 100 epochs)

Notifications You must be signed in to change notification settings

Prasanna28Devadiga/Cifar-10-

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Cifar-10-

Cifar-10 object recognition challenge with simple CNN (89% accuracy after 100 epochs). Techniques that helped increase accuracy (77%->89%):-

  1. Adding dropout layers
  2. Increasing the dropout value as we approach the final layer
  3. Adding batch-normalisation
  4. Training for a large number of epochs( Was initially training for 50 epochs)
  5. Using data augmentation

Notebook 2 - Uses Transfer Learning (Resnet-34) achieves about ~91% accuracy

About

Cifar-10 object recognition challenge with simple CNN (89% accuracy after 100 epochs)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published