Crowded Human Detection via an Anchor-pair Network
2020 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV)
Authors: Zhu, Jinguo; Yuan, Zejian; Zhang, Chong; Chi, Wanchao; Ling, Yonggen; Zhang, Shenghao
Abstract
This paper presents an anchor-pair network for crowded human detection, which can overcome and solve the difficulties caused by occlusion in crowded scenes. Specifically, we use a function-aware network structure to extract more distinctive and discriminative features for head and full-body respectively, and then a CNN module is also exploited to fuse the features by learning the correlations between head and full-body to reduce crowd errors. Meanwhile, a novel paired form for anchors, denoted as anchor-pair, is proposed to estimate the head regions and full-body regions simultaneously. Furthermore, a new ingenious Joint-NMS is introduced to perform on the detected head and full-body box pairs, which produces significant performance improvement in heavily occluded scenarios at tiny computational cost. Our anchor-pair network achieves a state-of-the-art result on the CrowdHuman dataset which reduces the MR-2 to 55.43%, achieving 11.59% relative improvement over our dataset baseline.
Prevention of leakage due to mouth opening through applying an oral shield device (Sominpax (TM)) during nasal CPAP therapy of patients with obstructive sleep apnea
SLEEP MEDICINE
Authors: Foellner, Sebastian; Guth, Patricia; Jorde, Ilka; Luecke, Eva; Ganzert, Christine; Stegemann-Koniszewski, Sabine; Schreiber, Jens
Abstract
Background/objective: The first line treatment for obstructive sleep apnea (OSA) is nasal continuous positive airway pressure (nCPAP), for which a variety of masks are available. While nasal masks (NM) are the first choice; oronasal masks (ONM) are also frequently used to prevent mouth dryness resulting from mouth opening. Our cross-sectional, prospective, randomized, un-blinded study addressed the efficacy of wearing an oral shield in addition to NM in preventing mouth leakage Methods: Patients with OSA and established therapy using NM and complaining about mouth dryness (n = 29) underwent three polysomnographies (PSGs) using NM, ONM or a nose mask in combination with an oral shield (NMS). Mask leakage was continuously documented and objective sleep quality was assessed. Results: There were significant differences in the apnea-hypopnea-index (AHI) between ONM (8.5/h; SD 6,7) and NM/nasal mask combined with oral shield device (NMS) (2.6/h; SD 2,3; 2.7/h; SD 2,6) (p < 0,05) as well as in leakage [ONM (39.7 l/min SD 12,4); NM (34.6 l/min SD 9,4); NMS (33.1 l/min SD 9,6)] (p = 0.011). Furthermore, analysis of sleep quality (NREM3) favored NM and NMS over ONM (p = 0.02). There were no significant differences between NM and NMS in any objective outcome. Conclusions: Our data consistently confirmed the NM as the first choice for continuous positive airway pressure (CPAP) therapy of OSA. Notably, we demonstrated a high potential of the oral shield for patients with mouth opening to achieve additional comfort and thereby possibly compliance, without affecting nCPAP therapy effectiveness. (C) 2019 Published by Elsevier B.V.