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 reconhecimento de atributos de pedestres (PAR) visa reconhecer uma série de atributos semânticos de uma pessoa, por exemplo, idade, sexo, o que desempenha um papel importante na videovigilância. Este artigo propõe uma rede convolucional de gráfico multicorrelação denominada MCGCN para PAR, que inclui um gráfico semântico, um gráfico visual e um gráfico de síntese. Construímos um gráfico semântico usando recursos de atributos com restrições semânticas. Uma convolução gráfica é empregada, com base no conhecimento prévio do conjunto de dados, para aprender a correlação semântica. Os recursos 2D são projetados em nós do gráfico visual e cada nó corresponde à região de recursos de cada grupo de atributos. A convolução do gráfico é então utilizada para aprender a correlação regional. Os nós do gráfico visual são conectados aos nós do gráfico semântico para formar um gráfico de síntese. No gráfico de síntese, as correlações regional e semântica são incorporadas entre si por meio de bordas entre gráficos, para orientar o aprendizado de cada um e para atualizar o gráfico visual e semântico, construindo assim a correlação semântica e regional. Nesta base, utilizamos uma melhor estratégia de ponderação de perdas, a suit_polyloss, para resolver o desequilíbrio dos conjuntos de dados de atributos de pedestres. Experimentos em três conjuntos de dados de referência mostram que a abordagem proposta atinge um desempenho de reconhecimento superior em comparação com as tecnologias existentes e atinge um desempenho de última geração.
Yang YU
Hebei University of Technology
Longlong LIU
Hebei University of Technology
Ye ZHU
Hebei University of Technology
Shixin CEN
Tianjin University of Traditional Chinese Medicine
Yang LI
Tianjin Academy of Agricultural Sciences
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
Yang YU, Longlong LIU, Ye ZHU, Shixin CEN, Yang LI, "MCGCN: Multi-Correlation Graph Convolutional Network for Pedestrian Attribute Recognition" in IEICE TRANSACTIONS on Information,
vol. E107-D, no. 3, pp. 400-410, March 2024, doi: 10.1587/transinf.2023EDP7134.
Abstract: Pedestrian attribute recognition (PAR) aims to recognize a series of a person's semantic attributes, e.g., age, gender, which plays an important role in video surveillance. This paper proposes a multi-correlation graph convolutional network named MCGCN for PAR, which includes a semantic graph, visual graph, and synthesis graph. We construct a semantic graph by using attribute features with semantic constraints. A graph convolution is employed, based on prior knowledge of the dataset, to learn the semantic correlation. 2D features are projected onto visual graph nodes and each node corresponds to the feature region of each attribute group. Graph convolution is then utilized to learn regional correlation. The visual graph nodes are connected to the semantic graph nodes to form a synthesis graph. In the synthesis graph, regional and semantic correlation are embedded into each other through inter-graph edges, to guide each other's learning and to update the visual and semantic graph, thereby constructing semantic and regional correlation. On this basis, we use a better loss weighting strategy, the suit_polyloss, to address the imbalance of pedestrian attribute datasets. Experiments on three benchmark datasets show that the proposed approach achieves superior recognition performance compared to existing technologies, and achieves state-of-the-art performance.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2023EDP7134/_p
Copiar
@ARTICLE{e107-d_3_400,
author={Yang YU, Longlong LIU, Ye ZHU, Shixin CEN, Yang LI, },
journal={IEICE TRANSACTIONS on Information},
title={MCGCN: Multi-Correlation Graph Convolutional Network for Pedestrian Attribute Recognition},
year={2024},
volume={E107-D},
number={3},
pages={400-410},
abstract={Pedestrian attribute recognition (PAR) aims to recognize a series of a person's semantic attributes, e.g., age, gender, which plays an important role in video surveillance. This paper proposes a multi-correlation graph convolutional network named MCGCN for PAR, which includes a semantic graph, visual graph, and synthesis graph. We construct a semantic graph by using attribute features with semantic constraints. A graph convolution is employed, based on prior knowledge of the dataset, to learn the semantic correlation. 2D features are projected onto visual graph nodes and each node corresponds to the feature region of each attribute group. Graph convolution is then utilized to learn regional correlation. The visual graph nodes are connected to the semantic graph nodes to form a synthesis graph. In the synthesis graph, regional and semantic correlation are embedded into each other through inter-graph edges, to guide each other's learning and to update the visual and semantic graph, thereby constructing semantic and regional correlation. On this basis, we use a better loss weighting strategy, the suit_polyloss, to address the imbalance of pedestrian attribute datasets. Experiments on three benchmark datasets show that the proposed approach achieves superior recognition performance compared to existing technologies, and achieves state-of-the-art performance.},
keywords={},
doi={10.1587/transinf.2023EDP7134},
ISSN={1745-1361},
month={March},}
Copiar
TY - JOUR
TI - MCGCN: Multi-Correlation Graph Convolutional Network for Pedestrian Attribute Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 400
EP - 410
AU - Yang YU
AU - Longlong LIU
AU - Ye ZHU
AU - Shixin CEN
AU - Yang LI
PY - 2024
DO - 10.1587/transinf.2023EDP7134
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E107-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2024
AB - Pedestrian attribute recognition (PAR) aims to recognize a series of a person's semantic attributes, e.g., age, gender, which plays an important role in video surveillance. This paper proposes a multi-correlation graph convolutional network named MCGCN for PAR, which includes a semantic graph, visual graph, and synthesis graph. We construct a semantic graph by using attribute features with semantic constraints. A graph convolution is employed, based on prior knowledge of the dataset, to learn the semantic correlation. 2D features are projected onto visual graph nodes and each node corresponds to the feature region of each attribute group. Graph convolution is then utilized to learn regional correlation. The visual graph nodes are connected to the semantic graph nodes to form a synthesis graph. In the synthesis graph, regional and semantic correlation are embedded into each other through inter-graph edges, to guide each other's learning and to update the visual and semantic graph, thereby constructing semantic and regional correlation. On this basis, we use a better loss weighting strategy, the suit_polyloss, to address the imbalance of pedestrian attribute datasets. Experiments on three benchmark datasets show that the proposed approach achieves superior recognition performance compared to existing technologies, and achieves state-of-the-art performance.
ER -