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
A detecção de pedestres é uma tarefa significativa na visão computacional. Nos últimos anos, tem sido amplamente utilizado em aplicações como sistemas de vigilância inteligentes e sistemas de condução automatizados. Embora tenha sido exaustivamente estudado na última década, a questão do manejo da oclusão ainda permanece sem solução. Uma ideia convincente é primeiro detectar partes do corpo humano e depois utilizar as informações das peças para estimar a existência dos pedestres. Muitas abordagens de detecção de pedestres baseadas em peças foram propostas com base nesta ideia. No entanto, na maioria dessas abordagens, a mineração de peças de baixa qualidade e a combinação desajeitada do detector de peças é um gargalo que limita o desempenho da detecção. Para eliminar o gargalo, propomos a Parte Discriminativa CNN (DP-CNN). Nossa abordagem tem duas contribuições principais: (1) Propomos um método de mineração de partes do corpo de alta qualidade baseado tanto nas características da camada convolucional quanto nas subclasses de partes do corpo. Os agrupamentos de partes minadas não são apenas discriminativos, mas também representativos, e podem ajudar a construir poderosos detectores de pedestres. (2) Propomos um novo método para combinar detectores de múltiplas partes. Convertemos os detectores parciais em uma camada intermediária de uma CNN e otimizamos todo o pipeline de detecção ajustando essa CNN. Em experimentos, mostra uma eficácia surpreendente de otimização e robustez no tratamento da oclusão.
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
Yu WANG, Cong CAO, Jien KATO, "Discriminative Part CNN for Pedestrian Detection" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 3, pp. 700-712, March 2022, doi: 10.1587/transinf.2021EDP7057.
Abstract: Pedestrian detection is a significant task in computer vision. In recent years, it is widely used in applications such as intelligent surveillance systems and automated driving systems. Although it has been exhaustively studied in the last decade, the occlusion handling issue still remains unsolved. One convincing idea is to first detect human body parts, and then utilize the parts information to estimate the pedestrians' existence. Many parts-based pedestrian detection approaches have been proposed based on this idea. However, in most of these approaches, the low-quality parts mining and the clumsy part detector combination is a bottleneck that limits the detection performance. To eliminate the bottleneck, we propose Discriminative Part CNN (DP-CNN). Our approach has two main contributions: (1) We propose a high-quality body parts mining method based on both convolutional layer features and body part subclasses. The mined part clusters are not only discriminative but also representative, and can help to construct powerful pedestrian detectors. (2) We propose a novel method to combine multiple part detectors. We convert the part detectors to a middle layer of a CNN and optimize the whole detection pipeline by fine-tuning that CNN. In experiments, it shows astonishing effectiveness of optimization and robustness of occlusion handling.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2021EDP7057/_p
Copiar
@ARTICLE{e105-d_3_700,
author={Yu WANG, Cong CAO, Jien KATO, },
journal={IEICE TRANSACTIONS on Information},
title={Discriminative Part CNN for Pedestrian Detection},
year={2022},
volume={E105-D},
number={3},
pages={700-712},
abstract={Pedestrian detection is a significant task in computer vision. In recent years, it is widely used in applications such as intelligent surveillance systems and automated driving systems. Although it has been exhaustively studied in the last decade, the occlusion handling issue still remains unsolved. One convincing idea is to first detect human body parts, and then utilize the parts information to estimate the pedestrians' existence. Many parts-based pedestrian detection approaches have been proposed based on this idea. However, in most of these approaches, the low-quality parts mining and the clumsy part detector combination is a bottleneck that limits the detection performance. To eliminate the bottleneck, we propose Discriminative Part CNN (DP-CNN). Our approach has two main contributions: (1) We propose a high-quality body parts mining method based on both convolutional layer features and body part subclasses. The mined part clusters are not only discriminative but also representative, and can help to construct powerful pedestrian detectors. (2) We propose a novel method to combine multiple part detectors. We convert the part detectors to a middle layer of a CNN and optimize the whole detection pipeline by fine-tuning that CNN. In experiments, it shows astonishing effectiveness of optimization and robustness of occlusion handling.},
keywords={},
doi={10.1587/transinf.2021EDP7057},
ISSN={1745-1361},
month={March},}
Copiar
TY - JOUR
TI - Discriminative Part CNN for Pedestrian Detection
T2 - IEICE TRANSACTIONS on Information
SP - 700
EP - 712
AU - Yu WANG
AU - Cong CAO
AU - Jien KATO
PY - 2022
DO - 10.1587/transinf.2021EDP7057
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2022
AB - Pedestrian detection is a significant task in computer vision. In recent years, it is widely used in applications such as intelligent surveillance systems and automated driving systems. Although it has been exhaustively studied in the last decade, the occlusion handling issue still remains unsolved. One convincing idea is to first detect human body parts, and then utilize the parts information to estimate the pedestrians' existence. Many parts-based pedestrian detection approaches have been proposed based on this idea. However, in most of these approaches, the low-quality parts mining and the clumsy part detector combination is a bottleneck that limits the detection performance. To eliminate the bottleneck, we propose Discriminative Part CNN (DP-CNN). Our approach has two main contributions: (1) We propose a high-quality body parts mining method based on both convolutional layer features and body part subclasses. The mined part clusters are not only discriminative but also representative, and can help to construct powerful pedestrian detectors. (2) We propose a novel method to combine multiple part detectors. We convert the part detectors to a middle layer of a CNN and optimize the whole detection pipeline by fine-tuning that CNN. In experiments, it shows astonishing effectiveness of optimization and robustness of occlusion handling.
ER -