Drone Imagery Object Detection - YOLO

The dataset utilized in this study was meticulously collected using drone video footage captured across seven distinct locations in Lithuania (Eastern Europe), each characterized by varying altitudes, camera angles, and weather conditions to ensure a diverse range of visual perspectives. The videos was taken from three cities. These are Vilnius, Šiauliai and Telšiai. The locations included a variety of road types such as intersections (T and + forms), slip lanes, normal roads, roundabouts, crossroads, parking areas, controlled T-form intersections, and dirt road intersections. Vilnius, being the capital city, presented a more complex traffic structure with multi-lane roads and heavy vehicle density. In contrast, Šiauliai and Telšiai provided a mix of suburban and rural environments with fewer vehicles. This geographical diversity ensures that the dataset captures a wide range of real-world traffic scenarios.

Labels: Car, Bus, Truck, Padestrian
Note: The study is about semi supervised learning. Only labeled data was provided here

本研究中所采用的数据集系通过精心搜集的无人机视频影像资料汇编而成，这些影像资料跨越立陶宛（东欧）七个不同地点，各地点海拔、摄像机角度及天气状况各异，以确保视觉视角的多样性。视频拍摄地点包括三个城市：维尔纽斯、希奥利乌和特拉希。所涉及的地点涵盖了多种道路类型，如交叉口（T型和+型）、匝道、普通道路、环形交叉路口、十字路口、停车区、受控T型交叉口和土路交叉口。作为首都，维尔纽斯展现了更为复杂的交通结构，拥有多车道道路和高车辆密度。相较之下，希奥利乌和特拉希则提供了郊区和乡村环境的混合景观，车辆数量较少。这种地理多样性确保了数据集能够捕捉到广泛的真实世界交通场景。

标签：汽车、公交车、卡车、行人

备注：本研究涉及半监督学习。此处仅提供了标记数据。