W2WDiff: Generalizing Underwater Diffusion Model via Unsupervised Underwater Conversion
W2WDiff is a diffusion-based underwater image enhancement (UIE) framework that integrates water-to-water domain aggregation with latent space operations to achieve superior generalization and enhancement performance. This design effectively addresses critical limitations of existing diffusion-based UIE methods, particularly their susceptibility to distribution shifts and semantic inconsistencies.

ViDA-UGC: Detailed Image Quality Analysis via Visual Distortion Assessment for UGC Images
To tackle the limitations of current explainable IQA methods in distinguishing UGC from AIGC and providing fine-grained quality analysis, we construct ViDA-UGC, the first large-scale UGC distortion assessment dataset with a distortion-oriented CoT annotation pipeline, and release ViDA-UGC-Bench as a reliable benchmark.

3D-UIR: 3D Gaussian for Underwater 3D Scene Reconstruction via Physics Based Appearance-Medium Decoupling
3D-UIR is an underwater NVS method that tackles water medium interference and floating artifacts in underwater scene representation through physics-based appearance-medium decoupling. Disentangles object appearance from water medium effects using tailored Gaussian modeling with appearance embeddings and distance-guided optimization.

Color Discrimination Multi-Scale Exposure Fusion for Underwater Image Enhancement
We propose CDEF, a reference-free multi-space pipeline that enhances underwater images by leveraging color discrimination and multi-scale fusion, achieving robust contrast, saturation, and sharpness improvements without artifacts or over-exposure.

Underwater Blurred Object Detection via Distraction Mining
UDMDet tackles underwater object detection challenges by exploring degradation mechanisms rather than image enhancement, mining target-background differences through distraction-aware feature fusion and task-aligned detection heads to progressively separate confusing features and improve detection performance in blurry underwater environments.

Multi-type Feature Perception and Refined Network for Spaceborne Infrared Ship Detection
MFPRN addresses satellite infrared ship detection challenges by enhancing information interaction during feature extraction through dual-branch feature fusion combining Fast Fourier convolution for comprehensive receptive fields and lightweight MLP for long-range dependencies, coupled with cascaded RPN to reduce false alarms in low SNR infrared imagery.

Real-World Underwater Image Texture Enhancement Based on Blurriness and Color Fusion
TEBCF focuses on underwater texture degradation caused by light scattering and absorption through multi-scale fusion of RGB contrast dehazing and CIELAB morphological enhancement, effectively recovering both global and local texture details while adaptively improving contrast, saturation, and sharpness.

Underwater Image Vision Enhancement Based on Contour Bougie Morphology
CBM addresses underwater texture degradation caused by forward scattering through multi-step morphological operations with dual-sized contour structure elements, enhancing underwater image visibility without requiring scene priors.