| Περίληψη: | Τhis Master Thesis targets upon extending the existing work of facial recognition algorithms regarding accuracy and efficiency and at proposing an online platform that can be used from the police forces towards effective real-time human recognition. The main features of the platform include a) inserting, editing and deleting user and criminal information and b) searching for criminals based on their picture through a livestream camera feed and identify them. The platform supports two different types of users: a) Police employees in the headquarters/precincts and b) Police Administrators, with a higher level of access and also responsible for database maintenance. Regarding the facial recognition algorithm, the approach uses and extends the Haar Cascade algorithm for real-time recognition, which is widely considered one for the most efficient and used algorithms for that cause. The development of the website was developed following the Model-View-Controller (MVC) architectural pattern, separating the platform in three different logical components. As for the criminal identification, it was made possible through image pattern recognition between the provided criminal’s image and snapshots of identified faces from the livestream feed. The platform includes a live feed section, accompanied with different options for video filters, enabling the user to select the best filter, depending on the relevant situation of the physical surroundings for better recognition results. At first, extensive research was conducted on both the state-of-the-art face recognition algorithm and the platforms dedicated for such causes and the capabilities and features they offer. After the initial research was conducted, the requirements definition followed, alongside with personas and scenarios development. Direct communication with relevant stakeholders who already have experience from these systems also occurred (e.g. through interviews). At some point, the platform was at an initial design stage (draft version), where the system should be meticulously studied and optimized towards its efficiency through a Heuristic Evaluation (HE) from usability experts and after gathering the relevant feedback and any other improvements to be implemented in the final version of the platform, the website entered its final stage. A questionnaire was disseminated in order to collect further feedback. After gathering the results and performing the necessary statistical analysis, we can proudly say that our approach was successful at its goals.
|
|---|
