Results 101 to 110 of about 9,649,421 (248)

Story2Board: A Training‐Free Approach for Expressive Visual Storytelling

Computer Graphics Forum, EarlyView.
Abstract We present Story2Board, a training‐free framework for expressive storyboard generation from natural language. Existing methods narrowly focus on subject identity, overlooking key aspects of visual storytelling such as spatial composition, background evolution, and narrative pacing.
D. Dinkevich, M. Levy, O. Avrahami, D. Samuel, D. Lischinski   +4 more
wiley   +1 more source

LoLA-SpecViT: Local attention SwiGLU vision transformer with LoRA for hyperspectral imaging

International Journal of Applied Earth Observations and Geoinformation
Hyperspectral image classification remains a challenging task due to the high dimensionality of spectral data, significant inter-band redundancy, and the limited availability of annotated samples.
Fadi Abdeladhim Zidi, Djamel Eddine Boukhari, Abdellah Zakaria Sellam, Abdelkrim Ouafi, Cosimo Distante, Salah Eddine Bekhouche, Abdelmalik Taleb-Ahmed   +6 more
doaj   +1 more source

Low-Rank Adaptation for Parameter-Efficient Fine-Tuning of Large Language Models

Modern natural language systems rely on large language models, whose sheer size makes full fine-tuning costly in computation, graphics processing unit (GPU) memory, and storage. Low-rank adaptation (LoRA) sidesteps most of that cost.
Shicheng Wei
core   +1 more source

Deep Residual Combiner: A Learned Fusion of Spatial, Temporal, and Multiscale Correlated Pixel Estimates

Computer Graphics Forum, EarlyView.
Abstract Correlation‐based rendering techniques continue to advance, and efficiently exploiting correlations between pixel estimates has become increasingly important. The deep combiner framework [BHHM20] allows us to fuse independent and correlated pixel estimates but focuses solely on spatial correlations.
W. Zhou, E. Hughes, T. Hachisuka
wiley   +1 more source

Generative Cutout Animation

Computer Graphics Forum, EarlyView.
Abstract Cutout animation is one of the earliest forms of animation, and to this day remains a popular technique featured in numerous films including Monty Python and South Park series. Most computer animation systems, however, focus on different styles, including cel animation, making cutout animation somewhat underexplored.
Ivan Puhachov, Noam Aigerman, Thibault Groueix, Mikhail Bessmeltsev   +3 more
wiley   +1 more source

Multiclass Diabetic Retinopathy Classification based on Low-Rank Adaptation

CLEI Electronic Journal
Diabetic retinopathy is a vision impairment associated with diabetes mellitus, a common medical condition. Retinographic imaging analysis of retinal fundus images is the predominant method of diagnosing this eye complication.
Sebastián Ferreria-Caballero, Diego P. Pinto-Roa, José Luis Vázquez Noguera, Jordan Ayala, Pastor Pérez-Estigarribia, Pedro E. Gardel-Sotomayor   +5 more
doaj   +1 more source

Single Line Drawing Generation via Semantics‐Driven Optimization

Computer Graphics Forum, EarlyView.
We present a method for automatically generating single‐line drawings in vector format, guided by a text prompt or an input image. Our approach leverages score distillation sampling to optimize the parameters of a uniform rational B‐spline (URBS) curve, ensuring that the drawing consists of a single continuous stroke by design.
Tanguy Magne, Alexandre Binninger, Ruben Wiersma, Olga Sorkine‐Hornung   +3 more
wiley   +1 more source

Tri‐LoRA: Spectral, Layer‐Selective, and Temporal LoRA Fusion for Preference‐Aligned Image Stylization

Computer Graphics Forum, EarlyView.
Abstract Personalized stylization based on subject and style LoRA fusion has emerged as a practical approach for controllable image generation. However, existing methods still struggle to balance subject consistency, style fidelity, and visual preference alignment, and many of them rely on additional training.
D. Zhang, Z. Zhu, Z. Peng, H. Dai, M. Han   +4 more
wiley   +1 more source

TENSLORA: TENSOR ALTERNATIVES FOR LOW-RANK ADAPTATION

International audienceLow-Rank Adaptation (LoRA) is widely used to efficiently adapt Transformers by adding trainable low-rank matrices to attention projections.
Bensaid, Reda, Lys, Jonathan, Leduc-Primeau, François, Gripon, Vincent, Marmoret, Axel   +4 more
core   +2 more sources

Efficient multi-tenant LoRA serving via SGMV-specific operator autotuning

Journal of King Saud University: Computer and Information Sciences
Low-Rank Adaptation (LoRA) has become a mainstream technique for fine-tuning large language models (LLMs) on downstream tasks due to its memory efficiency.
Chao Zhang, Gang Zhou, Teng Fu, Jicang Lu   +3 more
doaj   +1 more source