Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Ανίχνευση [προαιρετικά και ταξινόμηση] επικίνδυνων αντικειμένων σε χρόνο πραγματικό (real time) από δεδομένα video, με χρήση μεθόδων βαθιάς μάθησης

Tsavalis Nikolaos

Full record

---

Summary

This thesis will analyze computer vision and its potential in a wide range of applications and sciences through convolutional neural networks (CNN). The concepts of architectural structure and layers of the CNN will be developed as well as the techniques of training, pooling, normalization and activation that govern them. The generated code written in Python programming language and the technologies used to identify any weaponry will be given, first in static image and then in video. Through labeled data, it will be possible to identify the above weapon by indicating this type and any other weapon characteristics.

Available Files

Services