Brown Pedestrian Odometry Dataset (BPOD)

布朗行人航迹推算数据集（BPOD）是由布朗大学工程学院创建的一个用于评估头戴式行人视觉航迹推算算法的数据集。该数据集包含12个室内外多样化的地点，使用同步的全局和滚动快门立体相机捕捉。BPOD的特点是包含更多的图像模糊和自我旋转，这些在行人航迹推算中很常见。数据集的创建包括使用贴纸标记来定义行人的路径，并通过第三人的视频记录行人的位置。BPOD旨在解决头戴式摄像头在行人跟踪中的挑战，特别是在快速旋转和头部运动导致的图像模糊方面。

The Brown Pedestrian Dead Reckoning Dataset (BPOD) was developed by the School of Engineering at Brown University, serving as a benchmark dataset for evaluating head-mounted pedestrian visual dead reckoning algorithms. This dataset encompasses 12 diverse indoor and outdoor locations, with data captured using synchronized global and rolling shutter stereo cameras. A key feature of BPOD is its inclusion of a greater number of instances of image blur and self-rotation, both of which are frequently encountered scenarios in pedestrian dead reckoning tasks. The dataset construction process involved using sticker markers to define the pedestrian's traversal path, and recording the pedestrian's positions via video footage shot by a third individual. BPOD is specifically designed to address the challenges encountered by head-mounted cameras in pedestrian tracking, particularly those related to image blur induced by rapid rotation and head movements.