Ship detection and identification in video sequences

Abstract

This diploma thesis’s concern is a proposal of state-of-the-art techniques for object recognition, together with a description and a brief explanation of said techniques, implementation of a custom solution for ship recognition based on a chosen technique and its application on a real world problem. In the first part is a general introduction to the problematic of object and text detection. In the second part there is a detailed description of the chosen technique, which was used for the implementation of the custom solution. Then, in the next part, there is described the way of data processing and its use, which were made available for this project. Last but not least, there is an implementation proposal of the entire data engineering part, both for training and testing phases, together with implementation itself. And finally, in conclusion, a use of a chosen model on the prepared data, together with an objective evaluation and precision, effectivity and applicability appraisal for real world use.