EnvoDat
收藏资源简介:
EnvoDat数据集由蒙坦大学莱奥本分校创建,是一个大规模的多模态数据集,旨在为机器人提供在异构环境中的空间感知和语义推理能力。数据集包含26个序列,覆盖13个场景,数据量超过1.9TB,包含89K精细的多边形注释,适用于82个对象和地形类别。数据集在多种环境条件下采集,包括高光照、雾、雨和零可见度等。创建过程中,数据集经过多种格式的后处理,支持SLAM和监督学习算法的基准测试和多模态视觉模型的微调。EnvoDat数据集的应用领域包括环境适应性强的机器人自主性,特别是在极端挑战条件下。
The EnvoDat dataset, developed by Montanuniversität Leoben, is a large-scale multimodal dataset designed to equip robots with spatial awareness and semantic reasoning capabilities in heterogeneous environments. It contains 26 sequences spanning 13 distinct scenarios, with a total data volume exceeding 1.9 TB, and includes 89K fine-grained polygon annotations covering 82 object and terrain categories. The dataset was collected under diverse environmental conditions, including high illumination, fog, rain, and zero visibility scenarios. During its development, it underwent post-processing in multiple formats, enabling benchmarking of SLAM and supervised learning algorithms as well as fine-tuning of multimodal visual models. Application domains of the EnvoDat dataset include environment-adaptive robotic autonomy, particularly under extreme challenging conditions.

- 1EnvoDat: A Large-Scale Multisensory Dataset for Robotic Spatial Awareness and Semantic Reasoning in Heterogeneous Environments蒙坦大学莱奥本分校 · 2024年



