Journal of Infrared and Millimeter Waves, Volume. 40, Issue 4, 554(2021)
Infrared aircraft few-shot classification method based on meta learning
Figures & Tables(10)
Fig. 3. Comparison with different functions （a） improved with pool & add， （b） improved with conv & add， （c） improved with conv & concat
Fig. 5. Examples of datasets （a）mini-ImageNet dataset，（b）Infra-object dataset
Fig. 6. Test accuracy with different structures （a）5-way 1-shot， （b）5-way 5-shot
Fig. 7. Test accuracy of MAML and ours in different meta-train gradient steps （a）5-way 1-shot， （b）5-way 5-shot
Table 1. Accuracy and GFLOPs of different meta-train gradient steps on mini-ImageNet dataset
View tableView in ArticleTable 1. Accuracy and GFLOPs of different meta-train gradient steps on mini-ImageNet dataset
Method Meta-train gradient steps 5-way 1-shot 5-way 5-shot Accuracy /（%） GFLOPs Accuracy /（%） GFLOPs MLFC
（ours）
1 48.77±1.77 29.85 63.89±0.92 49.78 2 49.63±1.89 50.88 63.40±0.91 88.54 3 49.27±1.83 71.91 63.92±0.90 127.31 4 49.34±1.79 92.95 64.04±0.91 166.07 5 50.13±1.86 113.98 64.14±0.90 204.83 MAML［ 11 ］1 46.56+1.85 29.82 58.15+0.95 49.73 5 48.70±1.84 113.87 63.11±0.92 204.63
Table 2. Accuracy of different models on Infra-object dataset (%)
View tableView in ArticleTable 2. Accuracy of different models on Infra-object dataset (%)
Method Pretraining 8-way 1-shot 8-way 5-shot MAML No 74.62±0.99 90.25±0.38 MLFC （ours） No 78.58±0.97 91.12±0.37 MAML Yes 76.31±1.02 90.72±0.36 MLFC （ours） Yes 81.27±0.91 92.74±0.35
Table 3. Accuracy of different models on Infra-object dataset (%)
View tableView in ArticleTable 3. Accuracy of different models on Infra-object dataset (%)
Method 8-way 1-shot 8-way 5-shot Improved Relation Network［ 13 ］78.47±0.94 89.82±1.02 MLFC （ours） 81.27±0.91 92.74±0.35
Tools
Get Citation
Copy Citation Text
Rui-Min CHEN, Shi-Jian LIU, Zhuang MIAO, Fan-Ming LI. Infrared aircraft few-shot classification method based on meta learning[J]. Journal of Infrared and Millimeter Waves, 2021, 40(4): 554
Paper Information
Category: Research Articles
Received: Jul. 28, 2020
Accepted: --
Published Online: Sep. 9, 2021
The Author Email: Shi-Jian LIU (Shj_liu@ustc.edu)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20