航空学报2025，Vol.46Issue(24)：182-198,17.DOI:10.7527/S1000-6893.2025.31959
基于多阶段蒸馏的无人机图像时敏目标增量检测算法
Multi-stage distillation for incremental detection of time-sensitive targets in UAV images
摘要
Abstract
To address the problems of catastrophic forgetting,overfitting,and difficulty in adapting dense detector characteristics leading to limited detection accuracy faced by the current time-sensitive target class incremental object detection of UAV images,this paper proposes a time-sensitive target incremental detection algorithm based on multi-stage distillation.The proposed method includes three key modules:the continuous Wasserstein distance-based Inter-Class Distillation(WICD)module,the Prototype-Guided Intra-class Consistency Distillation(PGICD)module,and the Cross-prediction Adaptive Distillation(CAD)module.The WICD module captures the inter-class feature differences from the feature graphs and semantic query vectors using Gaussian distributions with continuous Wasserstein dis-tances to enhance the inter-class discriminability.The PGICD module enhances the inter-class discriminative proper-ties by minimizing the high-level semantic query of the instances in both the teacher's network and student's network,and the low-level feature graphs with the prototype differences to achieve effective intra-class feature transfer and en-hance intra-class consistency.The CAD module optimizes the cross-prediction distillation process by dynamically ad-justing the distillation weights of the classification and regression branches,mitigating the problem of catastrophic for-getting in incremental learning,and improving the model's detection accuracy in complex scenarios.Experimental re-sults on SIMD and MAR20 datasets show that the proposed method performs well in all types of one-step and multi-step incremental scenarios,and the Average Precision(AP)is significantly improved compared with traditional meth-ods.The AP is as high as 70.8%in the incremental scenario of SIMD dataset with 8+7 classes,with an absolute gap of 1.7%and a relative gap of 2.3%from the upper limit.The results also show that the AP is as high as 60.2%in the incremental scenario of MAR20 dataset with 10+10 classes,with an absolute gap of 2.3%and a relative gap of 3.6%from the upper limit.In addition,the effectiveness of each module is verified by ablation experiments,which effectively improves the incremental detection performance of time-sensitive targets in UAV images.关键词
增量目标检测/知识蒸馏/无人机图像/时敏目标检测/交叉预测蒸馏Key words
incremental object detection/knowledge distillation/UAV images/time-sensitive target detec-tion/cross-prediction distillation分类
航空航天引用本文复制引用
CHENG Zhenhao,YANG Xiaogang,LU Ruitao,ZHANG Tao,WANG Siyu..基于多阶段蒸馏的无人机图像时敏目标增量检测算法[J].航空学报,2025,46(24):182-198,17.基金项目
陕西省重点研发计划(2024CY2-GJHX-42) （2024CY2-GJHX-42）
国家自然科学基金(62276274) （62276274）
陕西省三秦英才特殊支持计划(2024-SQ-001) Key Research and Development Program of Shaanxi Province(2024CY2-GJHX-42) （2024-SQ-001）
National Natural Science Foundation of China(62276274) （62276274）
Shaanxi SanqinElite Special Support Program(2024-SQ-001) （2024-SQ-001）
