To address the issue of low aircraft classification accuracy caused by the scarcity of high-precision labeled data, which hinders effective model training, an aircraft classification and recognition method based on transfer learning and motion feature fusion is proposed. Firstly, differences in motion characteristics of different types of aircraft are analyzed. Subsequently, the RGB algorithm is employed to transform motion-differentiated features into trajectory images that encapsulate both trajectory shape information and target motion attributes. Finally, for scenarios with limited sample data, a transfer learning strategy is introduced, fine-tuning and optimizing a pre-trained ResNet model to accomplish aircraft model classification and recognition. Experimental results demonstrate that the proposed method achieves an aircraft model classification accuracy of 75.6% on small-sample datasets, outperforming baseline models by an improvement of 11 to 15.3 percentage points.
PEDROCHED S, AMIGOD, GARCÍAJ, et al. Architecture for trajectory-based fishing ship classification with AIS data[J]. Sensors, 2020,20(13):No.3782.
[2]
HUANGH G, HONGF, LIUJ, et al. FVID: fishing vessel type identification based on VMS trajectories[J]. Journal of Ocean University of China, 2019,18(2):403-412.
[3]
ENDOY, TODAH, NISHIDAK, et al. Deep feature extraction from trajectories for transportation mode estimation[C]//Proceedings of the Advances in Knowledge Discovery and Data Mining. Cham, Switzerland: Springer, 2016:54-66.
[4]
KONTOPOULOSI, MAKRISA, TSERPESK. A deep learning streaming methodology for trajectory classification[J]. ISPRS International Journal of Geo-Information, 2021,10(4):250.
[5]
CHENX, LIUY C, ACHUTHANK, et al. A ship movement classification based on Automatic Identification System (AIS) data using Convolutional Neural Network[J]. Ocean Engineering, 2020,218:No.108182.
[6]
LUS C, XIAY. Dual supervised autoencoder based trajectory classification using enhanced spatio-temporal information[J]. IEEE Access, 2020,8:173918-173932.
CORTÉSE, SÁNCHEZS. Deep learning transfer with AlexNet for chest X-ray COVID-19 recognition[J]. IEEE Latin America Transactions, 2021,19(6):944-951.
[14]
YENDURIG, RAMALINGAMM, SELVIG C, et al. GPT (generative pre-trained transformer): a comprehensive review on enabling technologies, potential applications, emerging challenges, and future directions[J]. IEEE Access, 2024,12:54608-54649.
[15]
LIUY, HUT G, ZHANGH R, et al. iTransformer: inverted transformers are effective for time series forecasting[DB/OL]. (2023-10-10)[2025-12-22].
[16]
WUH X, HUT G, LIUY,et al. TimesNet: temporal 2D-variation modeling for general time series analysis[DB/OL]. (2022-10-05)[2025-12-22].
[17]
LIANGY X, OUYANGK, WANGY W,et al. TrajFormer: efficient trajectory classification with transformers[C]//Proceedings of the 31st ACM International Conference on Information & Knowledge Management. Atlanta GA USA. New York: ACM, 2022:1229-1237.