🏢 Nanjing University of Science and Technology

Progressive Exploration-Conformal Learning (PECL) revolutionizes sparsely annotated object detection in aerial images by adaptively selecting high-quality pseudo-labels, overcoming limitations of exis…

Boosting label distribution learning accuracy and efficiency, this research proposes using ternary labels instead of binary labels to predict label distributions, thus enhancing annotation efficiency …

MambaLLIE: a novel implicit Retinex-aware low-light enhancer using a global-then-local state space, significantly outperforms existing CNN and Transformer-based methods.

Dynamic Rebalancing Contrastive Learning with Dual Reconstruction (2DRCL) pre-training significantly boosts object detection accuracy, especially for underrepresented classes.

IMAGPose: A unified framework generating high-fidelity person images from single or multiple source images & poses, addressing existing methods’ limitations.

Researchers dynamically integrate contrastive and supervised learning to overcome the modality imbalance problem in multimodal classification, significantly improving model performance.

DoFIT: A novel domain-aware framework significantly reduces catastrophic forgetting in federated instruction tuning by finely aggregating overlapping weights and using a proximal perturbation initiali…

DCDepth achieves state-of-the-art monocular depth estimation by progressively predicting depth in the frequency domain via DCT, capturing local correlations and global context effectively.

NerPE: continuous heatmap regression via implicit neural representation resolves the accuracy-limiting quantization errors in human pose estimation, achieving sub-pixel precision.