GitXplorerGitXplorer
r

Conscience

public
0 stars
0 forks
0 issues

Commits

List of commits on branch master.
Unverified
3dbddd93728914c62f2199b06707f145cbe1cd06

Updated README

rrharish101 committed 7 years ago
Unverified
6a4e784e07797d84565c72c232004f9bf4623f1e

Removed unnecesary files

rrharish101 committed 7 years ago
Unverified
25c03f58251868c7708593b5562de522f2ea3a13

Added files

rrharish101 committed 7 years ago
Unverified
4873bc5b8ee0aefa87387f15e5a70563d9dc16f5

Initial commit

rrharish101 committed 7 years ago

README

The README file for this repository.

Conscience

This is my code for question 1 of the Machine Learning event called Conscience held in Takneek '17 at IIT Kanpur.

Contents

  • q1.py: The main file to train a fine-tuned VGG16 model on the given dataset.
  • 256_ObjectClassification.zip: The dataset for question 1, a set of images classified into 257 categories (not 256). Download from https://drive.google.com/drive/folders/0B8ZTgwx4fdLua1RIM3RxNG5BOUk (if link is broken, then email me).
  • q1_vgg_mod.gz0x: Split parts of a tarball containing the final fine-tuned VGG16 model, a Keras Sequential model. Concatenate the split parts into one .tar.gz file and then extract it.
  • q1_dense_top.h5: The Keras Sequential model for the fully-connected layers to be used instead of VGG16's fully-connected layers.
  • q1_dense_top_weights.h5: The Keras Sequential weights for the fully-connected layers to be used instead of VGG16's fully-connected layers.
  • All other files are numpy arrays stored for ease of access when debugging or running the code again.