PulseAugur / Brief
EN
LIVE 13:52:30

Brief

last 24h
[1/1] 223 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. PRISM: Rethinking Atmospheric Scattering Reconstruction as a Unified Understanding and Restoration Model for Real-world Dehazing

    Researchers have introduced PRISM, a new framework for reconstructing clear images from hazy real-world scenes. PRISM utilizes a physically structured approach called Proximal Scattering Atmosphere Reconstruction (PSAR) to jointly restore the clear scene and atmospheric scattering variables, enhancing interpretability. To address the lack of paired data, the system includes an online haze synthesis pipeline and a Selective Self-Distillation Adaptation (SSDA) scheme for unpaired learning, enabling self-refinement based on perceptual targets and scattering understanding. Experiments show PRISM achieves competitive results on real-world dehazing benchmarks. AI