Systém pro efektivní rozšiřování databází určených k detekci objektů v obraze

Abstract

This thesis deals with a system for the efficient expansion of databases intended for object detection in images. The theoretical part describes the issues of automatic object detection, compares models based on convolutional neural networks, and thoroughly analyses the requirements for databases designed for automatic object detection. In the practical part, a system is proposed that enables the efficient extension of a database for vehicle detection, accompanied by a complete user manual.