FG-DSA: Frequency-Guided Depthwise Separable Attention for Efficient Dense Prediction
cs.CV
Submitted: 2026-11-07
Updated: 2026-11-15
Terminology
Sources
- FGA: Fourier-Guided Attention Network for Crowd Count Estimation
- Dynamic Convolution: Attention over Convolution Kernels
- Frequency Dynamic Convolution for Dense Image Prediction
- Deep Residual Learning for Image Recognition
- Dynamic Sparse Attention: Access Patterns and Architecture
- Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs