Daimler Pedestrian Detection Benchmark Dataset

涉及大型培训和测试集。训练集包含15.560行人样本 (以48x96分辨率截取的图像) 和6744不包含行人的额外完整图像，用于提取负样本。该测试集包含一个独立的序列，其中包含21.790多个带有56.492行人标签 (完全可见或部分遮挡) 的图像，这些图像在经过城市交通的27分钟车程期间从车辆捕获，分辨率为VGA (640x480，未压缩)。因此，数据集是现实的，并且在发布时 (8.5 Gb) 比其他数据集大大约一个数量级。

This dataset includes large-scale training and test sets. The training set contains 15,560 pedestrian samples (images cropped at 48×96 resolution) and 6,744 additional complete images without pedestrians for negative sample extraction. The test set consists of an independent sequence with over 21,790 images bearing 56,492 pedestrian annotations (fully visible or partially occluded). These images were captured from a moving vehicle during a 27-minute drive through urban traffic, with a VGA resolution of 640×480 and uncompressed. As a result, this dataset is highly realistic, and was approximately an order of magnitude larger than other datasets at the time of its release, with a total size of 8.5 Gb.

搜集汇总

背景与挑战

背景概述

Daimler Pedestrian Detection Benchmark Dataset是一个用于行人检测的大规模数据集，包含15,560个行人样本和6,744个非行人图像，以及21,790个测试图像。数据集图像从车辆捕获，具有现实场景特点，总规模约为8.5 Gb。