FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism

Ying Liu; Luyao Geng; Hao Yu; Zhijie Xu

doi:10.1145/3488933.3489000

FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism

Ying Liu, Luyao Geng, Hao Yu, Zhijie Xu

Xi'an University of Posts and Telecommunications

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The current target detection algorithm can be competent for most of the detection tasks. However, improving the detection accuracy of small targets is difficult due to the small target occupy less pixels and the feature extraction is hard to achieve. To address this problem, the per-pixel target detection algorithm FCOS is adapted in this research, and the widely used ResNet50 is implemented as the algorithm backbone, by adjusting the size of the input image and the composition of the loss function. The CBAM hybrid attention mechanism is applied into the shallow features and high-level features corresponding to the bottom pyramid, then the feature pyramid is constructed to achieve the purpose of multi-scale target detection. The comparison and ablation experiments show that the original FCOS target detection model can improve the detection accuracy by about 3.7% and the small target detection accuracy by about 2% on the MS-COCO dataset.

Original language	English
Title of host publication	AIPR 2021 - 2021 4th International Conference on Artificial Intelligence and Pattern Recognition
Publisher	Association for Computing Machinery (ACM)
Pages	50-55
Number of pages	6
ISBN (Electronic)	9781450384087
DOIs	https://doi.org/10.1145/3488933.3489000
Publication status	Published - 24 Sept 2021
Event	4th International Conference on Artificial Intelligence and Pattern Recognition - Virtual, Online, China Duration: 24 Sept 2021 → 26 Sept 2021 Conference number: 4 https://dl.acm.org/doi/proceedings/10.1145/3488933

Conference

Conference	4th International Conference on Artificial Intelligence and Pattern Recognition
Abbreviated title	AIPR 2021
Country/Territory	China
City	Virtual, Online
Period	24/09/21 → 26/09/21
Internet address	https://dl.acm.org/doi/proceedings/10.1145/3488933

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

Access to Document

10.1145/3488933.3489000Licence: Unspecified

Cite this

@inproceedings{b1a423d743334ebe8a72cb64fe956fae,

title = "FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism",

abstract = "The current target detection algorithm can be competent for most of the detection tasks. However, improving the detection accuracy of small targets is difficult due to the small target occupy less pixels and the feature extraction is hard to achieve. To address this problem, the per-pixel target detection algorithm FCOS is adapted in this research, and the widely used ResNet50 is implemented as the algorithm backbone, by adjusting the size of the input image and the composition of the loss function. The CBAM hybrid attention mechanism is applied into the shallow features and high-level features corresponding to the bottom pyramid, then the feature pyramid is constructed to achieve the purpose of multi-scale target detection. The comparison and ablation experiments show that the original FCOS target detection model can improve the detection accuracy by about 3.7\% and the small target detection accuracy by about 2\% on the MS-COCO dataset. ",

keywords = "Anchor-Free, Convolutional neural network, Hybrid Attention Mechanism, Object detection",

author = "Ying Liu and Luyao Geng and Hao Yu and Zhijie Xu",

note = "Publisher Copyright: {\textcopyright} 2021 ACM.; 4th International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2021 ; Conference date: 24-09-2021 Through 26-09-2021",

year = "2021",

month = sep,

day = "24",

doi = "10.1145/3488933.3489000",

language = "English",

pages = "50--55",

booktitle = "AIPR 2021 - 2021 4th International Conference on Artificial Intelligence and Pattern Recognition",

publisher = "Association for Computing Machinery (ACM)",

address = "United States",

url = "https://dl.acm.org/doi/proceedings/10.1145/3488933",

}

Liu, Y, Geng, L, Yu, H & Xu, Z 2021, FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism. in AIPR 2021 - 2021 4th International Conference on Artificial Intelligence and Pattern Recognition. Association for Computing Machinery (ACM), pp. 50-55, 4th International Conference on Artificial Intelligence and Pattern Recognition, Virtual, Online, China, 24/09/21. https://doi.org/10.1145/3488933.3489000

FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism. / Liu, Ying; Geng, Luyao; Yu, Hao et al.
AIPR 2021 - 2021 4th International Conference on Artificial Intelligence and Pattern Recognition. Association for Computing Machinery (ACM), 2021. p. 50-55.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism

AU - Liu, Ying

AU - Geng, Luyao

AU - Yu, Hao

AU - Xu, Zhijie

N1 - Conference code: 4

PY - 2021/9/24

Y1 - 2021/9/24

N2 - The current target detection algorithm can be competent for most of the detection tasks. However, improving the detection accuracy of small targets is difficult due to the small target occupy less pixels and the feature extraction is hard to achieve. To address this problem, the per-pixel target detection algorithm FCOS is adapted in this research, and the widely used ResNet50 is implemented as the algorithm backbone, by adjusting the size of the input image and the composition of the loss function. The CBAM hybrid attention mechanism is applied into the shallow features and high-level features corresponding to the bottom pyramid, then the feature pyramid is constructed to achieve the purpose of multi-scale target detection. The comparison and ablation experiments show that the original FCOS target detection model can improve the detection accuracy by about 3.7% and the small target detection accuracy by about 2% on the MS-COCO dataset.

AB - The current target detection algorithm can be competent for most of the detection tasks. However, improving the detection accuracy of small targets is difficult due to the small target occupy less pixels and the feature extraction is hard to achieve. To address this problem, the per-pixel target detection algorithm FCOS is adapted in this research, and the widely used ResNet50 is implemented as the algorithm backbone, by adjusting the size of the input image and the composition of the loss function. The CBAM hybrid attention mechanism is applied into the shallow features and high-level features corresponding to the bottom pyramid, then the feature pyramid is constructed to achieve the purpose of multi-scale target detection. The comparison and ablation experiments show that the original FCOS target detection model can improve the detection accuracy by about 3.7% and the small target detection accuracy by about 2% on the MS-COCO dataset.

KW - Anchor-Free

KW - Convolutional neural network

KW - Hybrid Attention Mechanism

KW - Object detection

UR - https://www.scopus.com/pages/publications/85125836912

U2 - 10.1145/3488933.3489000

DO - 10.1145/3488933.3489000

M3 - Conference contribution

AN - SCOPUS:85125836912

SP - 50

EP - 55

BT - AIPR 2021 - 2021 4th International Conference on Artificial Intelligence and Pattern Recognition

PB - Association for Computing Machinery (ACM)

T2 - 4th International Conference on Artificial Intelligence and Pattern Recognition

Y2 - 24 September 2021 through 26 September 2021

ER -

FCOS Small Target Detection Algorithm Combined with Multi-Layer Hybrid Attention Mechanism

Abstract

Conference

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this