MVP-Occ
收藏资源简介:
MVP-Occ是一个用于密集行人场景的多视角行人占用预测的新型合成数据集,由韩国科学技术院创建。该数据集包含五个大规模场景,模拟真实世界环境,提供了详细的行人使用体素结构的表示,并伴随丰富的语义场景理解标签。数据集的创建过程利用了CARLA模拟器,生成了多样化的城市交通场景。MVP-Occ数据集的应用领域主要集中在城市监控系统中的行人识别和场景理解,旨在解决多视角行人检测中的占用状态预测问题。
MVP-Occ is a novel synthetic dataset for multi-view pedestrian occupancy prediction in dense pedestrian scenarios, created by the Korea Advanced Institute of Science and Technology. It contains five large-scale scenes that simulate real-world environments, providing detailed voxel-based representations of pedestrians and accompanying rich semantic scene understanding annotations. The dataset was developed using the CARLA simulator to generate diverse urban traffic scenarios. The main application areas of the MVP-Occ dataset focus on pedestrian recognition and scene understanding in urban surveillance systems, aiming to solve the occupancy state prediction problem in multi-view pedestrian detection.

- 1Multi-View Pedestrian Occupancy Prediction with a Novel Synthetic Dataset韩国科学技术院 · 2024年



