Data analytics algorithms for multi-dimensional datasets: an image classifier to recognize different species of flowers
Αλγόριθμοι ανάλυσης πολυδιάστατων δεδομένων
KeywordsΑλγόριθμοι ; Τεχνητή νοημοσύνη ; Μηχανική μάθηση ; Βαθιά μάθηση ; Algorithms ; Artificial intelligence ; Machine learning ; Deep learning
Hundreds of flowers exist on earth, consisting an integral part of all livings not only for the aesthetic aspect but also for human life in many areas such as medical science, industry and environment. It is necessary to set up a database for flower documentation by determining an effective mean to identify the species to which they belong even from a smartphone application. As Artificial Intelligence algorithms (Neural Networks) are more and more incorporated into everyday applications, developing such an image classifier by creating a deep learning model trained on hundreds of thousands of images would drive as part of the overall application architecture. In this work, there is developed and trained an image classifier for recognizing 102 distinct species of flowers utilizing a certain type of machine learning algorithm called Convolutional Neural Networks, resulting in very good performance on a variety of experiments, achieving up to 94% accuracy.