SimilarityImage

The similarity comparison of flower images, conducted using keras, opencv, and tensorflow, has been performed.

Table of Contents

1. Dataset
2. Model
3. How to run
4. Results

Dataset

The dataset consists of 102 images of flowers. Here is an example of one of the images:

Model

The objective of this project is to provide the top 10 most similar images to a given flower image using the A Discriminative Feature Learning Approach for Deep Face Recognition paper. We utilized a combination of the Center loss and Cross Entropy techniques. Finally, we evaluated the model using the ROC-AUC metric.

How to run

1. The "Prepare" folder contains the data preparation scripts to format the dataset in a similar manner to the LFW dataset. It is not necessary to run these scripts separately.
2. Run ImageSimlarityFinally (3).ipynb Notebook

Results

Query Image	10 similar images