EGSA-PT:Edge-Guided Spatial Attention with Progressive Training for Monocular Depth Estimation and Segmentation of Transparent Objects

TL;DR

This paper introduces EGSA, a boundary-aware fusion mechanism, and a progressive training strategy that together improve depth estimation and segmentation of transparent objects without requiring ground-truth depth during training.

Contribution

The paper proposes a novel edge-guided spatial attention fusion method and a multi-modal progressive training approach for better transparent object perception.

Findings

EGSA improves depth accuracy over state-of-the-art methods.

The progressive training strategy eliminates the need for ground-truth depth.

Significant improvements in transparent regions on Syn-TODD and ClearPose benchmarks.

Abstract

Transparent object perception remains a major challenge in computer vision research, as transparency confounds both depth estimation and semantic segmentation. Recent work has explored multi-task learning frameworks to improve robustness, yet negative cross-task interactions often hinder performance. In this work, we introduce Edge-Guided Spatial Attention (EGSA), a fusion mechanism designed to mitigate destructive interactions by incorporating boundary information into the fusion between semantic and geometric features. On both Syn-TODD and ClearPose benchmarks, EGSA consistently improved depth accuracy over the current state of the art method (MODEST), while preserving competitive segmentation performance, with the largest improvements appearing in transparent regions. Besides our fusion design, our second contribution is a multi-modal progressive training strategy, where learning transitions from edges derived from RGB images to edges derived from predicted depth images. This approach allows the system to bootstrap learning from the rich textures contained in RGB images, and then switch to more relevant geometric content in depth maps, while it eliminates the need for ground-truth depth at training time. Together, these contributions highlight edge-guided fusion as a robust approach capable of improving transparent object perception.