About

GOOSE is the German Outdoor and Offroad Dataset and is a 2D & 3D semantic segmentation dataset framework with a modern dataset specification. The original specifcation came along with a main GOOSE dataset containing 10000 frames. Another 5000 frames from additional platforms for finetuning and transfer learning is referred to as GOOSE-Ex dataset.

In contrast to existing datasets like Cityscapes or BDD100K, the focus is on unstructured off-road environments, similar to RELLIS-3D. GOOSE supports different robot platforms and applications in the fields of mobile robotics and deep learning.

GOOSE contains annotated point clouds, segmentation masks from RGB and near-infrared (NIR) images, as well as accurate localization.

The initial release in 2024 contains annotated data from MuCAR-3. The GOOSE-Ex dataset in the same framework with annotated data of ALICE and Spot will follow in Q1/2025.

Data Privacy

We take your privacy seriously! All image raw data has been processed to blur faces and other personally identifiable information such as number plates. However, if you find yourself or personal belongings in this dataset and feel unwell about it, please contact us and we will postprocess or remove the respective data.