Current state-of-the-art object-centric models use slots and attention-based routing for binding. However, this class of models has several conceptual limitations: the number of slots is hardwired; ...

Abstract: Recent research has shown that deep learning models are likely to make incorrect predictions even when exposed to minor perturbations. To address this, training models on adversarial ...

InstructSAM is a training-free framework for Instruction-Oriented Object Counting, Detection, and Segmentation (InstructCDS). We construct EarthInstruct, an InstructCDS benchmark for remote sensing.

Abstract: This paper presents a novel approach that leverages two models to integrate features from numerous unlabeled images, addressing the challenge of semi-supervised salient object detection ...