EVP: Enhanced Visual Perception using Inverse Multi-Attentive Feature Refinement and Regularized Image-Text Alignment

TL;DR

This paper introduces EVP, an advanced network architecture that enhances visual perception by refining features and aligning images with text, achieving state-of-the-art results in depth estimation and referring segmentation tasks.

Contribution

The paper proposes the IMAFR module and a novel image-text alignment method, significantly improving feature learning and extraction in the Stable Diffusion backbone.

Findings

State-of-the-art depth estimation on NYU Depth v2 and KITTI datasets.

Significant IoU improvement in referring segmentation on RefCOCO.

Enhanced feature learning capabilities demonstrated through comprehensive experiments.

Abstract

This work presents the network architecture EVP (Enhanced Visual Perception). EVP builds on the previous work VPD which paved the way to use the Stable Diffusion network for computer vision tasks. We propose two major enhancements. First, we develop the Inverse Multi-Attentive Feature Refinement (IMAFR) module which enhances feature learning capabilities by aggregating spatial information from higher pyramid levels. Second, we propose a novel image-text alignment module for improved feature extraction of the Stable Diffusion backbone. The resulting architecture is suitable for a wide variety of tasks and we demonstrate its performance in the context of single-image depth estimation with a specialized decoder using classification-based bins and referring segmentation with an off-the-shelf decoder. Comprehensive experiments conducted on established datasets show that EVP achieves state-of-the-art results in single-image depth estimation for indoor (NYU Depth v2, 11.8% RMSE improvement over VPD) and outdoor (KITTI) environments, as well as referring segmentation (RefCOCO, 2.53 IoU improvement over ReLA). The code and pre-trained models are publicly available at https://github.com/Lavreniuk/EVP.