Zhiwen Tao, Fu Niu. Image Segmentation Method of Military Personnel in Multiple Complex Environments Based on U-Net[J]. Laser & Optoelectronics Progress, 2022, 59(8): 0810005
- Laser & Optoelectronics Progress
- Vol. 59, Issue 8, 0810005 (2022)
Fig. 1. Architecture of DESI-U-Net
Fig. 2. Structure of improved DAC module
Fig. 3. Structure of serial attention mechanism
Fig. 4. Six example images of complex environments in MECD. (a) Jungle; (b) city ruins; (c) mountain field; (d) rainforest; (e) desert; (f) snowfield
Fig. 5. Partial segmentation results of four networks in 6 complex environments. (a) Original image; (b) ground truth; (c) segmentation results of FCN-8s; (d) segmentation results of SegNet; (e) segmentation results of U-Net; (f) segmentation results of DESI-U-Net
|
Table 1. Parameter amounts of network with different type of convolution
|
Table 2. Camouflage type information in MECD
|
Table 3. mIoU of different methods
|
Table 4. Parameter amounts of different methods
|
Table 5. Results of ablation experiment
Set citation alerts for the article
Please enter your email address
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20