The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
O rastreamento humano múltiplo é amplamente utilizado em vários campos, como marketing e vigilância. A abordagem típica associa resultados de detecção humana entre quadros consecutivos usando os recursos e caixas delimitadoras (posição+tamanho) dos humanos detectados. Alguns métodos usam uma câmera omnidirecional para cobrir uma área mais ampla, mas a troca de ID geralmente ocorre em associação com detecções devido aos seguintes dois fatores: i) O recurso é afetado negativamente porque a caixa delimitadora inclui muitas regiões de fundo quando um ser humano é capturado de uma posição oblíqua. ângulo. ii) A posição e o tamanho mudam drasticamente entre quadros consecutivos porque a métrica de distância não é uniforme em uma imagem omnidirecional. Neste artigo, propomos um novo método que rastreia humanos com precisão com uma métrica de associação para imagens omnidirecionais. O método proposto possui dois pontos principais: i) Para extração de características, introduzimos a retificação local, o que reduz o efeito das regiões de fundo na caixa delimitadora. ii) Para cálculo de distância, descrevemos as posições em um sistema de coordenadas mundial onde a métrica de distância é uniforme. Nos experimentos, confirmamos que a precisão de rastreamento de múltiplos objetos (MOTA) melhorou 3.3 no conjunto de dados LargeRoom e melhorou 2.3 no conjunto de dados SmallRoom.
Hitoshi NISHIMURA
KDDI Research, Inc.,Nagoya University
Naoya MAKIBUCHI
KDDI Research, Inc.
Kazuyuki TASAKA
KDDI Research, Inc.
Yasutomo KAWANISHI
KDDI Research, Inc.,Nagoya University
Hiroshi MURASE
KDDI Research, Inc.,Nagoya University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Hitoshi NISHIMURA, Naoya MAKIBUCHI, Kazuyuki TASAKA, Yasutomo KAWANISHI, Hiroshi MURASE, "Multiple Human Tracking Using an Omnidirectional Camera with Local Rectification and World Coordinates Representation" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 6, pp. 1265-1275, June 2020, doi: 10.1587/transinf.2019MVP0009.
Abstract: Multiple human tracking is widely used in various fields such as marketing and surveillance. The typical approach associates human detection results between consecutive frames using the features and bounding boxes (position+size) of detected humans. Some methods use an omnidirectional camera to cover a wider area, but ID switch often occurs in association with detections due to following two factors: i) The feature is adversely affected because the bounding box includes many background regions when a human is captured from an oblique angle. ii) The position and size change dramatically between consecutive frames because the distance metric is non-uniform in an omnidirectional image. In this paper, we propose a novel method that accurately tracks humans with an association metric for omnidirectional images. The proposed method has two key points: i) For feature extraction, we introduce local rectification, which reduces the effect of background regions in the bounding box. ii) For distance calculation, we describe the positions in a world coordinate system where the distance metric is uniform. In the experiments, we confirmed that the Multiple Object Tracking Accuracy (MOTA) improved 3.3 in the LargeRoom dataset and improved 2.3 in the SmallRoom dataset.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019MVP0009/_p
Copiar
@ARTICLE{e103-d_6_1265,
author={Hitoshi NISHIMURA, Naoya MAKIBUCHI, Kazuyuki TASAKA, Yasutomo KAWANISHI, Hiroshi MURASE, },
journal={IEICE TRANSACTIONS on Information},
title={Multiple Human Tracking Using an Omnidirectional Camera with Local Rectification and World Coordinates Representation},
year={2020},
volume={E103-D},
number={6},
pages={1265-1275},
abstract={Multiple human tracking is widely used in various fields such as marketing and surveillance. The typical approach associates human detection results between consecutive frames using the features and bounding boxes (position+size) of detected humans. Some methods use an omnidirectional camera to cover a wider area, but ID switch often occurs in association with detections due to following two factors: i) The feature is adversely affected because the bounding box includes many background regions when a human is captured from an oblique angle. ii) The position and size change dramatically between consecutive frames because the distance metric is non-uniform in an omnidirectional image. In this paper, we propose a novel method that accurately tracks humans with an association metric for omnidirectional images. The proposed method has two key points: i) For feature extraction, we introduce local rectification, which reduces the effect of background regions in the bounding box. ii) For distance calculation, we describe the positions in a world coordinate system where the distance metric is uniform. In the experiments, we confirmed that the Multiple Object Tracking Accuracy (MOTA) improved 3.3 in the LargeRoom dataset and improved 2.3 in the SmallRoom dataset.},
keywords={},
doi={10.1587/transinf.2019MVP0009},
ISSN={1745-1361},
month={June},}
Copiar
TY - JOUR
TI - Multiple Human Tracking Using an Omnidirectional Camera with Local Rectification and World Coordinates Representation
T2 - IEICE TRANSACTIONS on Information
SP - 1265
EP - 1275
AU - Hitoshi NISHIMURA
AU - Naoya MAKIBUCHI
AU - Kazuyuki TASAKA
AU - Yasutomo KAWANISHI
AU - Hiroshi MURASE
PY - 2020
DO - 10.1587/transinf.2019MVP0009
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 2020
AB - Multiple human tracking is widely used in various fields such as marketing and surveillance. The typical approach associates human detection results between consecutive frames using the features and bounding boxes (position+size) of detected humans. Some methods use an omnidirectional camera to cover a wider area, but ID switch often occurs in association with detections due to following two factors: i) The feature is adversely affected because the bounding box includes many background regions when a human is captured from an oblique angle. ii) The position and size change dramatically between consecutive frames because the distance metric is non-uniform in an omnidirectional image. In this paper, we propose a novel method that accurately tracks humans with an association metric for omnidirectional images. The proposed method has two key points: i) For feature extraction, we introduce local rectification, which reduces the effect of background regions in the bounding box. ii) For distance calculation, we describe the positions in a world coordinate system where the distance metric is uniform. In the experiments, we confirmed that the Multiple Object Tracking Accuracy (MOTA) improved 3.3 in the LargeRoom dataset and improved 2.3 in the SmallRoom dataset.
ER -