Real-Time Object Detection and Tracking for Unmanned Aerial Vehicles Based on Convolutional Neural Networks

Yang, Shao-Yu; Cheng, Hsu-Yung; Yu, Chih-Chang

doi:10.3390/electronics12244928

Cited by 5 publications

(2 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To improve the reliability of the self-supervised training, we employ a variable weight λ cons for L cons . The λ cons shows the ratio of ŷS,T with combined confidence C comb greater than the predefined threshold value C th , as presented in Equation (9).…”

Section: Training Algorithmmentioning

confidence: 99%

“…Object detection [1][2][3] aims to locate and classify the targets in the given image, which has received significant attention in computer vision recently. With the emergence of deep feed-forward architectures [4][5][6], modern data-driven detection methods [1][2][3][7][8][9][10] lead to considerable improvements in many applications, including security surveillance, autonomous driving, and so on. However, those achievements are obtained only when test data and training data maintain the same distribution.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AdvMix: Adversarial Mixing Strategy for Unsupervised Domain Adaptive Object Detection

Chen,

Lv,

Dai

et al. 2024

Electronics

View full text Add to dashboard Cite

Recent object detection networks suffer from performance degradation when training data and test data are distinct in image styles and content distributions. In this paper, we propose a domain adaptive method, Adversarial Mixing (AdvMix), where the label-rich source domain and unlabeled target domain are jointly trained by the adversarial feature alignment and a self-training strategy. To diminish the style gap, we design the Adversarial Gradient Reversal Layer (AdvGRL), containing a global-level domain discriminator to align the domain features by gradient reversal, and an adversarial weight mapping function to enhance the stability of domain-invariant features by hard example mining. To eliminate the content gap, we introduce a region mixing self-supervised training strategy where a region of the target image with the highest confidence is selected to merge with the source image, and the synthesis image is self-supervised by the consistency loss. To improve the reliability of self-training, we propose a strict confidence metric combining both object and bounding box uncertainty. Extensive experiments conducted on three benchmarks demonstrate that AdvMix achieves prominent performance in terms of detection accuracy, surpassing existing domain adaptive methods by nearly 5% mAP.

show abstract

Section: Training Algorithmmentioning

confidence: 99%