Object detection from DIOR dataset with YOLOv5 and YOLOv8

Authors

Karam Mawas
Raissa Magana Ugarte

Dataset

The dataset used in our experiments is DIOR DATASET (a brief description of the dataset: 23463 images, each of 800x800 pixels. Its main objective is to be used to develop data-driven methods for detecting multiple different objects from a satellite imagery.)

Model

The models employed are YOLOv5 and YOLOv8, sizes nano, small and medium.

Results

We obtained the following results: