Urban100
PulseAugur coverage of Urban100 — every cluster mentioning Urban100 across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Graphene: Lightweight Transformer Enhances Image Super-Resolution
Researchers have introduced Graphene, a novel lightweight Transformer architecture designed for image super-resolution. Graphene incorporates two key refinement mechanisms: an Ising Refinement Block for local feature re…
-
Study questions value of expensive image demosaicing methods
A new analysis explores the cost-effectiveness of different image demosaicing methods, which reconstruct full-color images from single-sensor data. Researchers found that the quality gain from expensive edge-aware metho…
-
New protocol and dataset evaluate super-resolution artifacts by perceptual impact
Researchers have introduced SR-Prominence, a new crowdsourced protocol and dataset designed to evaluate artifacts in image super-resolution (SR) methods based on perceptual impact. The protocol defines artifact prominen…
-
New AI models boost image super-resolution efficiency and quality
Two new research papers introduce advanced image super-resolution techniques. CEM-TUDASR, a Transformer-based framework, enhances low-resolution images from wireless capsule endoscopy without requiring paired training d…
-
New Wavelet-Domain Model Achieves Efficient JPEG Restoration
Researchers have developed a new generative model for JPEG image restoration that operates efficiently in the wavelet domain. This model, named MeanFlow, utilizes a Haar transform instead of a VAE encoder-decoder and in…
-
New GB-LSR method offers faster, continuous image reconstruction
Researchers have introduced GB-LSR, a novel local spectral image representation designed for continuous image reconstruction and super-resolution. This method partitions images into patches, with each patch containing c…
-
New GB-LSR method offers faster, more efficient image reconstruction
Researchers have introduced GB-LSR, a novel local spectral image representation designed for efficient continuous image reconstruction and super-resolution. This method partitions images into patches, each using Fourier…
-
QuantSR+ framework boosts low-bit quantized image super-resolution
Researchers have developed QuantSR+, a novel framework designed to enhance the performance of image super-resolution models that utilize low-bit quantization. This approach addresses the significant performance degradat…