Pixel Level
Pixel-level analysis focuses on understanding and manipulating individual pixels within images and videos, aiming for precise object segmentation and detailed scene understanding. Current research emphasizes developing efficient and robust methods for pixel-level tasks, particularly in weakly or unsupervised settings, often leveraging transformer architectures, diffusion models, and foundation models like SAM and CLIP to reduce reliance on extensive manual annotation. This work has significant implications for diverse applications, including remote sensing, medical image analysis, and computer vision generally, by enabling more accurate and automated image interpretation and manipulation.
Papers
PRIMA: Multi-Image Vision-Language Models for Reasoning Segmentation
Muntasir Wahed, Kiet A. Nguyen, Adheesh Sunil Juvekar, Xinzhuo Li, Xiaona Zhou, Vedant Shah, Tianjiao Yu, Pinar Yanardag, Ismini Lourentzou
A Super-pixel-based Approach to the Stable Interpretation of Neural Networks
Shizhan Gong, Jingwei Zhang, Qi Dou, Farzan Farnia