Flowers Classification Project (unfinished)

In this repository can be found half-term project for my BigData Methods and Tools university classes. It is an image classification problem in which we are given different photos of flowers and are supposed to write a programme which will be able to classify them by specified species. To deal with this problem I implemented the Convolutional Neural Network using Keras and TensorFlow librares. I also wanted to see how various classifiers would dealt with that type of data, so I used SVM and decision trees with different features extractions methods. To make this task even more demanding, apart from taking flowers photos from Flowers Dataset, I decided to search for more data using images gathering programs. Then I had to check wheter they are valid or not, usually one by one. I came across with plenty of problems - from low quality to photos where flower was barely visible. However, the most difficult was to get rid of photos which contained species which were other than the one I was looking for. Exhausting and it took ages, but appriopriate and reliable data is the key to success. Every single step I had to perform is described in flowers_classification_manual.ipynb file. I have also wanted to play with the MobileNetV2 CNN and results of this can be found here. I have also implemented SVM with HOG feature extraction method - SVM (RBF) with HOG features extraction. If you check them, you will notice that there are a very few lines of code actually - each function which I wrote and called is stored in functions.py. To conclude, I encourage you to check these files out! If you have some questions, it would be a pleasure for me to try to answer them ;)