Transfer Learning using VGG16

---

About:

This is a deep learning project based on transfer learning. Here i have used VGG16 as a pre-trained model and trained a dataset of flower_images on it.

---

Motive:

Motive behind this project was to improve the accuracy of the model by implementing transfer learning technique.

---

Libraries Used:

Numpy
Keras
Tensorflow
Matplotlib
glob

---

Process:

I re-trained the last layer of VGG16 model with training data consisting of flower images.

The model was re-trained on google colab's gpu and is giving an accuracy of > 80% in just 5 epochs. Given this data set also consisits of landscapes and garbage images , it is a massive improvement, further training couldn't be done due to lack of better hardware.

---

About the dataset:

It is a subset of 102flower data set provide by Oxford.It consists of 5 categories of 5 categories of flowers: Daisy, Dandelion, roses, tulips and Suflower.

---

Biblography:

1. Blogs: Aqeel Anwar Jason Brownlee
2. Youtube: Krish Naik Deeplearning.ai
   Special thanks for Krish Naik who help me clear the concept and also provide a generic code template that is applicable for all transfer learning projects.