Real-Time Maritime Semaphore Sign Detection System

In times of distress, the maritime signalling system using semaphore signs play a crucial role in facilitating communication and ensuring the safety and efficiency of maritime operations. Traditional methods that have been used over the years of semaphore sign communication rely on manual interpretation, which can be subjective, time-consuming, and prone to human error. The interpretation can also vary depending on the observer's skill and experience, leading to inconsistencies in message comprehension. The proposed system aims to provide an integrated solution to enable communication with the help of detection of semaphore signs. It utilizes convolutional neural networks (CNNs) for classification of semaphore signs in real-time. The dataset consisting of semaphore signs for six letters of the English language i.e, ‘A’, ‘E’, ‘F’, ‘O’, ‘S’, ‘V’ is created and used to train the model. The CNN model trained on this dataset showed an accuracy of 99.20% on the train set and 98.87% on the test set while detecting semaphore signs in real time.