FARCLUSS: fuzzy adaptive rebalancing and contrastive uncertainty learning for semi-supervised semantic segmentation [0.03%]
基于模糊自适应重标定和对比不确定性学习的半监督语义分割方法
Ebenezer Tarubinga,Jenifer Kalafatovich,Seong-Whan Lee
Ebenezer Tarubinga
Semi-supervised semantic segmentation (SSSS) faces persistent challenges in effectively leveraging unlabeled data, including ineffective utilization of pseudo-labels, the exacerbation of class imbalance biases, and the neglect of prediction...
GraphMinNet: Learning dependencies in graphs with light complexity minimal architecture [0.03%]
图神经网络轻量级架构设计及依赖关系学习方法研究
Md Atik Ahamed,Andrew Cheng,Qiang Ye et al.
Md Atik Ahamed et al.
Graph Neural Networks (GNNs) have demonstrated remarkable success in various applications. Despite their success, traditional GNNs are inherently limited by their local message-passing paradigm and often struggle to capture long-range depen...
DFMamba: Difference-enhanced and adaptive frequency-aware Mamba for change detection [0.03%]
差异增强的自适应频率感知马玛巴变化检测方法
Panli Yuan,Xingbiao Zhao,Renlong Hang
Panli Yuan
Recently, the Mamba architecture has demonstrated outstanding performance in natural language processing and general vision tasks, and has been rapidly extended to the field of Remote Sensing Change Detection (RSCD). However, existing Mamba...
DECHRL: Empowerment-driven delay-aware causal hierarchical reinforcement learning [0.03%]
基于赋能的延迟感知因果分层强化学习(DECHRL)
Chenran Zhao,Dianxi Shi,Haotian Wang et al.
Chenran Zhao et al.
Many real-world tasks involve delayed effects, where the outcomes of actions emerge after varying time lags. Existing delay-aware reinforcement learning methods often rely on state augmentation, prior knowledge of delay distributions, or ac...
MEMC: Masked modeling with efficient and minimal contrastive learning for self-supervised skeleton-based action recognition [0.03%]
基于高效极简对比学习的自监督姿态动作识别方法MEMC
Yingfei Wu,Wenming Cao,Xinpeng Yin
Yingfei Wu
Unsupervised 3D skeleton-based action recognition offers advantages in terms of robustness and computational efficiency. However, prevailing paradigms, such as masked skeleton modeling (MSM) and contrastive learning (CL), have inherent limi...
Enhanced multi-agent interaction modeling with differential transformers for trajectory prediction [0.03%]
基于微分变压器的增强型多智能体交互模型在轨迹预测中的应用
Minghui Wei,Haimin Zhang,Stuart Perry et al.
Minghui Wei et al.
Predicting the future trajectories of multiple agents is intrinsically challenging due to complex dynamic agent-scene interactions, diverse motion patterns, and the inherent uncertainty of real-world behaviors. Transformer-based architectur...
Identity-style preserving image composition with domain-adaptive diffusion [0.03%]
基于领域自适应扩散的身份风格保持图像合成方法
Qi Si,Zhao Zhang,Yang Zhao et al.
Qi Si et al.
In generative image compositing, effectively achieving cross-domain and same-domain composition while balancing semantic fidelity and visual coherence remains a key challenge. Specifically, current research faces three critical limitations:...
HFFNet: Hierarchical feature fusion network for lightweight RGB-T semantic segmentation [0.03%]
基于分层特征融合的轻量型RGB-T语义分割网络(HFFNet)
Xiaoling Wu,Xiaofei Zhou,Liuxin Bao et al.
Xiaoling Wu et al.
In recent years, the integration of RGB and thermal infrared (TIR) images has shown great potential in advancing semantic segmentation, particularly under challenging conditions such as low-light environments and cluttered backgrounds. Howe...
Ke Jin,Yuhu Cheng,Yi Kong et al.
Ke Jin et al.
Implicit Q-learning, a prominent approach within the implicit value learning (IVL) paradigm, mitigates distribution shift in offline reinforcement learning (offline RL) by avoiding out-of-distribution action queries. However, current IVL me...
CRFormer: Multi-scale contrastive regularization transformer for visible watermark removal [0.03%]
CRFormer:用于去除可见水印的多尺度对比正则化变压器模型
Yun Hao,Jiaju Wu,Ruizhou Sun et al.
Yun Hao et al.
Visible watermarks are widely used for image copyright protection, but their removal remains a challenging restoration task due to the diversity of watermarks in color, scale, transparency, and spatial distribution. Existing methods predomi...