VGG16-Transfer-Learning-TF

This repository contains the transfer learning over VGG16 architecture implemented in Tensorflow, using tf.slim

Author: Ravin Kumar

Libraries required:

• Tensorflow version : 1.12

• OpenCV version : 2

Directory Architecture:

|
|-transfer_learning.py 
|
|-inference.py 
|
|-vgg_16.ckpt (link is provided to download pretrained weights)
|
|-time (stores log file for overall time, start time, auto-shutdown time)
|
|-saved (trained weights)
|
|-log (stores visual graph for tensorboard)
|
|-data
   |-Class1 (contains *.jpg for class1)
   |
   |-Class2 (contains *.jpg for class2)
   |
   |-Class3 (contains *.jpg for class3)
   |
   |-Class4 (contains *.jpg for class4)

Training code

python3 transfer_learning.py

Inference code

python3 inference.py

Download VGG16.ckpt file from Google Drive

Download

Note: This work can be used freely after providing citation and/or deserved credits to this work. Please use it at your own risk !!!