Romanian Coins Recognition and Sum Counting System From Image Using TensorFlow and Keras

The daily necessity for automation in simple and repetitive jobs, as well as in industry, has resulted in rapid technological advancement. Several organizations choose to combine robotic tools for physical jobs, as well as Artificial Intelligence (AI), Machine Learning (ML), and Neural Network (NN) -based applications for optimizing the flow of work, minimizing human error, and executing chores faster, safer, and more reliably. Simple tasks that require a little time every day can be done automatically by creating a simple AI-based app, conserving us a lot of time in the long run. Because time equals money, a technology that saves time may be extremely important and necessary for many people and businesses. Counting money, particularly coins, is a simple chore that causes many firms to waste several minutes each day and that may be conveniently completed by an application. Even though many physical machines can do this task, there are several drawbacks: they can be expensive if the device's accuracy is great, they are huge, not portable, and are viewed as a waste of money for small enterprises. Such counting machines are simply replaced with an application that recognizes the value of the coins and counts the exact value. A coin-counting application can be useful for calculating the total number of coins inside a deposit account from home, or for calculating coins at the end of a shift in commerce. The objective of this work is to create an application that calculates the total worth of Romanian coins from an input image. The developed application is divided into two components: coin recognition and coin classification. The end application should be able to detect currency in an image and distinguish between coins worth 10 Bani and coins worth 50 Bani.