Cross-view geo-localization, crucial for 6G applications like drone navigation and 3D reconstruction, is hampered by the inherent challenges of achieving feature consistency across diverse viewpoints and capturing globally discriminative information. Existing methods often focus on extracting discriminative global geographic features. Nevertheless, the substantial appearance variations across different viewpoints can hinder effective localization. Additionally, the consistency between dual-perspective features is frequently overlooked, resulting in mismatches and inaccurate localization. To tackle these issues, we propose a novel Supervised Contrastive Orthogonal Fusion framework (SCOF) tailored for cross-view geo-localization. SCOF comprises a Supervised Contrastive Learning Module (SCM) and an Orthogonal Fusion Module (OFM). Supervised Contrastive Learning Module (SCM) leverages semantic labels to enforce consistent feature representations across different perspectives, effectively mitigating the issue of misalignment. The integrated Orthogonal Fusion Module (OFM) extracts orthogonal global components from local features, fostering a synergistic fusion of local and global information for enhanced discriminative capability. Extensive evaluations on benchmark datasets (University-1652, SUES-200 and DenseUAV) demonstrate that SCOF achieves state-of-the-art performance in drone-view target localization and drone navigation tasks, surpassing existing approaches in accuracy and robustness. By effectively bridging the gap between local and global feature representations while ensuring cross-view consistency, SCOF offers a significant advancement in the field of cross-view geo-localization.

Face representation in the wild is extremely hard due to the large scale face variations. Some deep convolutional neural networks (CNNs) have been developed to learn discriminative feature by designing properly margin-based losses, which perform well on easy samples but fail on hard samples. Although some methods mainly adjust the weights of hard samples in training stage to improve the feature discrimination, they overlook the distribution property of feature. It is worth noting that the miss-classified hard samples may be corrected from the feature distribution view. To overcome this problem, this paper proposes the hard samples guided optimal transport (OT) loss for deep face representation, OTFace in short. OTFace aims to enhance the performance of hard samples by introducing the feature distribution discrepancy while maintaining the performance on easy samples. Specifically, we embrace triplet scheme to indicate hard sample groups in one mini-batch during training. OT is then used to characterize the distribution differences of features from the high level convolutional layer. Finally, we integrate the margin-based-softmax (e.g. ArcFace or AM-Softmax) and OT together to guide deep CNN learning. Extensive experiments were conducted on several benchmark databases. The quantitative results demonstrate the advantages of the proposed OTFace over state-of-the-art methods.