Vehicle re-identification (ReID) is a challenging task in intelligent transportation, and urban surveillance systems due to its complications in camera viewpoints, vehicle scales, and environmental conditions. Recent transformer-based approaches have shown impressive performance by utilizing global dependencies, these models struggle with aspect ratio distortions and may overlook fine-grained local attributes crucial for distinguishing visually similar vehicles. We introduce a framework based on Swin Transformers that addresses these challenges by implementing three components. First, to improve feature robustness and maintain vehicle proportions, our Aspect Ratio-Aware Swin Transformer (AR-Swin) preserve the native ratio via letterbox, uses a non-square (16 $\times$ 8) patch-embedding stem, and keeps fixed 7 $\times$ 7 token windows. Second, we introduce a Dynamic Feature Fusion Network (DFFNet) that adaptively integrates global Swin features with local attribute embeddings; such as color and vehicle type enabling more discriminative representations. Third, our Regional Attention Blocks incorporate regional masks into the transformer’s windowed attention mechanism, effectively highlighting critical details like manufacturer logos or lights. On VeRi-776, we obtain 82.55 mAP, 97.26 Rank-1 and 99.23 Rank-5, and on VehicleID we obtain 91.8 Rank-1 and 97.75 Rank-5. The design is drop-in for Swin backbones and emphasizes robustness without increasing architectural complexity. Code: https://github.com/sft110/Swinvreid.

Visual diagnosis of skin cancer is challenging due to subtle inter-class similarities, variations in skin texture, the presence of hair, and inconsistent illumination. Deep learning models have shown promise in assisting early detection, yet their performance is often limited by the severe class imbalance present in dermoscopic datasets. This paper proposes CANNSkin, a skin cancer classification framework that integrates a convolutional autoencoder with latent-space oversampling to address this imbalance. The autoencoder is trained to reconstruct lesion images, and its latent embeddings are used as features for classification. To enhance minority-class representation, the Synthetic Minority Oversampling Technique (SMOTE) is applied directly to the latent vectors before classifier training. The encoder and classifier are first trained independently and later fine-tuned end-to-end. On the HAM10000 dataset, CANNSkin achieves an accuracy of $93.01\mathrm{\%}$, a macro-F1 of $88.54\mathrm{\%}$, and an ROC–AUC of $98.44\mathrm{\%}$, demonstrating strong robustness across ten test subsets. Evaluation on the more complex ISIC 2019 dataset further confirms the model’s effectiveness, where CANNSkin achieves 94.27% accuracy, 93.95% precision, 94.09% recall, and 99.02% F1-score, supported by high reconstruction fidelity (PSNR 35.03 dB, SSIM 0.86). These results demonstrate the effectiveness of our proposed latent-space balancing and fine-tuned representation learning as a new benchmark method for robust and accurate skin cancer classification across heterogeneous datasets.