The thermal imaging pedestrian segmentation system has excellent performance in different illumination conditions, but it has some drawbacks(e.g., weak pedestrian texture information, blurred object boundaries). Meanwhile, high-performance large models have higher latency on edge devices with limited computing performance. To solve the above problems, in this paper, we propose a real-time thermal infrared pedestrian segmentation method. The feature extraction layers of our method consist of two paths. Firstly, we utilize the lossless spatial downsampling to obtain boundary texture details on the spatial path. On the context path, we use atrous convolutions to improve the receptive field and obtain more contextual semantic information. Then, the parameter-free attention mechanism is introduced at the end of the two paths for effective feature selection, respectively. The Feature Fusion Module (FFM) is added to fuse the semantic information of the two paths after selection. Finally, we accelerate method inference through multi-threading techniques on the edge computing device. Besides, we create a high-quality infrared pedestrian segmentation dataset to facilitate research. The comparative experiments on the self-built dataset and two public datasets with other methods show that our method also has certain effectiveness. Our code is available at https://github.com/mcjcs001/LEIPNet.
Shangdong LIU
Nanjing University of Posts and Telecommunications
Chaojun MEI
Nanjing University of Posts and Telecommunications
Shuai YOU
Nanjing University of Posts and Telecommunications
Xiaoliang YAO
Nanjing University of Posts and Telecommunications
Fei WU
Nanjing University of Posts and Telecommunications
Yimu JI
Nanjing University of Posts and Telecommunications
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.
Copy
Shangdong LIU, Chaojun MEI, Shuai YOU, Xiaoliang YAO, Fei WU, Yimu JI, "A Lightweight and Efficient Infrared Pedestrian Semantic Segmentation Method" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 9, pp. 1564-1571, September 2023, doi: 10.1587/transinf.2022EDP7217.
Abstract: The thermal imaging pedestrian segmentation system has excellent performance in different illumination conditions, but it has some drawbacks(e.g., weak pedestrian texture information, blurred object boundaries). Meanwhile, high-performance large models have higher latency on edge devices with limited computing performance. To solve the above problems, in this paper, we propose a real-time thermal infrared pedestrian segmentation method. The feature extraction layers of our method consist of two paths. Firstly, we utilize the lossless spatial downsampling to obtain boundary texture details on the spatial path. On the context path, we use atrous convolutions to improve the receptive field and obtain more contextual semantic information. Then, the parameter-free attention mechanism is introduced at the end of the two paths for effective feature selection, respectively. The Feature Fusion Module (FFM) is added to fuse the semantic information of the two paths after selection. Finally, we accelerate method inference through multi-threading techniques on the edge computing device. Besides, we create a high-quality infrared pedestrian segmentation dataset to facilitate research. The comparative experiments on the self-built dataset and two public datasets with other methods show that our method also has certain effectiveness. Our code is available at https://github.com/mcjcs001/LEIPNet.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022EDP7217/_p
Copy
@ARTICLE{e106-d_9_1564,
author={Shangdong LIU, Chaojun MEI, Shuai YOU, Xiaoliang YAO, Fei WU, Yimu JI, },
journal={IEICE TRANSACTIONS on Information},
title={A Lightweight and Efficient Infrared Pedestrian Semantic Segmentation Method},
year={2023},
volume={E106-D},
number={9},
pages={1564-1571},
abstract={The thermal imaging pedestrian segmentation system has excellent performance in different illumination conditions, but it has some drawbacks(e.g., weak pedestrian texture information, blurred object boundaries). Meanwhile, high-performance large models have higher latency on edge devices with limited computing performance. To solve the above problems, in this paper, we propose a real-time thermal infrared pedestrian segmentation method. The feature extraction layers of our method consist of two paths. Firstly, we utilize the lossless spatial downsampling to obtain boundary texture details on the spatial path. On the context path, we use atrous convolutions to improve the receptive field and obtain more contextual semantic information. Then, the parameter-free attention mechanism is introduced at the end of the two paths for effective feature selection, respectively. The Feature Fusion Module (FFM) is added to fuse the semantic information of the two paths after selection. Finally, we accelerate method inference through multi-threading techniques on the edge computing device. Besides, we create a high-quality infrared pedestrian segmentation dataset to facilitate research. The comparative experiments on the self-built dataset and two public datasets with other methods show that our method also has certain effectiveness. Our code is available at https://github.com/mcjcs001/LEIPNet.},
keywords={},
doi={10.1587/transinf.2022EDP7217},
ISSN={1745-1361},
month={September},}
Copy
TY - JOUR
TI - A Lightweight and Efficient Infrared Pedestrian Semantic Segmentation Method
T2 - IEICE TRANSACTIONS on Information
SP - 1564
EP - 1571
AU - Shangdong LIU
AU - Chaojun MEI
AU - Shuai YOU
AU - Xiaoliang YAO
AU - Fei WU
AU - Yimu JI
PY - 2023
DO - 10.1587/transinf.2022EDP7217
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2023
AB - The thermal imaging pedestrian segmentation system has excellent performance in different illumination conditions, but it has some drawbacks(e.g., weak pedestrian texture information, blurred object boundaries). Meanwhile, high-performance large models have higher latency on edge devices with limited computing performance. To solve the above problems, in this paper, we propose a real-time thermal infrared pedestrian segmentation method. The feature extraction layers of our method consist of two paths. Firstly, we utilize the lossless spatial downsampling to obtain boundary texture details on the spatial path. On the context path, we use atrous convolutions to improve the receptive field and obtain more contextual semantic information. Then, the parameter-free attention mechanism is introduced at the end of the two paths for effective feature selection, respectively. The Feature Fusion Module (FFM) is added to fuse the semantic information of the two paths after selection. Finally, we accelerate method inference through multi-threading techniques on the edge computing device. Besides, we create a high-quality infrared pedestrian segmentation dataset to facilitate research. The comparative experiments on the self-built dataset and two public datasets with other methods show that our method also has certain effectiveness. Our code is available at https://github.com/mcjcs001/LEIPNet.
ER -