【摘要】 在行人检测任务中，最主要的是能够快速且精准识别图像或视频中的行人。由于在密集的场景中行人之间存在相互遮挡，导致目前行人检测算法普遍存在漏检和误检问题。为了解决上述问题，文章提出了一种基于改进Soft-NMS算法的多实例行人检测方法。为了减小由于行人之间重叠导致检测器无法做出准确预测，引入多实例检测方法来使检测器做出相对准确的预测。并且在Soft-NMS算法的基础上进行改进，提出了Set-Soft-NMS算法来减少行人检测中的漏检和误检问题。使用公开的行人检测数据集上对文章所提出的方法进行实验，结果表明该文提出方法的性能要优于其他主流的行人检测算法。更多还原

【Abstract】 In the task of pedestrian detection, the most important thing is to be able to quickly and accurately identify pedestrians in images or videos.Due to mutual occlusion among pedestrians in dense scenes, current pedestrian detection algorithms generally have problems of missed detection and false detection.In order to solve the above problems, this paper proposes a multi-instance pedestrian detection method based on the improved Soft-NMS algorithm.In order to reduce the inability of the detector to make accurate predictions due to the overlap between pedestrians, a multi-instance detection method is introduced to make the detectors make relatively accurate predictions.And based on the Soft-NMS algorithm, the Set-Soft-NMS algorithm is improved to reduce the missed detection and false detection problems in pedestrian detection.The method proposed in this paper is tested on the public pedestrian detection data set, and the results show that the performance of the method proposed in this paper is better than other mainstream pedestrian detection algorithms.更多还原