用深度图边缘对齐提升弱监督语义分割性能
Depth Edge Alignment Loss: DEALing with Depth in Weakly Supervised Semantic Segmentation
- 利用图像级标签生成像素级标签，结合深度信息增强监督
- 在PASCAL VOC、MS COCO等数据集上提升5.439至16.416点mIoU
- 适用于机器人系统，可与其它损失函数组合使用
自主机器人系统在新领域应用需大量昂贵的像素级密集标注来训练鲁棒的语义分割模型。本文提出一种模型无关的深度边缘对齐损失（Depth Edge Alignment Loss, DEAL），用于提升不同数据集上的弱监督语义分割性能。该方法通过图像级监督生成像素级语义标签，避免高成本标注。尽管弱监督在传统计算机视觉中已有广泛研究，本工作引入了机器人系统中常见且易获取的像素级深度信息作为补充监督。实验表明，所提方法在多个数据集和模型上均显著提升分割性能，尤其在PASCAL VOC、MS COCO验证集及HOPE静态接入子集上，平均交并比（mIoU）分别提升+5.439、+1.274和+16.416点。代码已公开于https://github.com/DTU-PAS/DEAL。
原文摘要 · Abstract (English)
Autonomous robotic systems applied to new domains require an abundance of expensive, pixel-level dense labels to train robust semantic segmentation models under full supervision. This study proposes a model-agnostic Depth Edge Alignment Loss to improve Weakly Supervised Semantic Segmentation models across different datasets. The methodology generates pixel-level semantic labels from image-level supervision, avoiding expensive annotation processes. While weak supervision is widely explored in traditional computer vision, our approach adds supervision with pixel-level depth information, a modality commonly available in robotic systems. We demonstrate how our approach improves segmentation performance across datasets and models, but can also be combined with other losses for even better performance, with improvements up to +5.439, +1.274 and +16.416 points in mean Intersection over Union on the PASCAL VOC / MS COCO validation, and the HOPE static onboarding split, respectively. Our code is made publicly available at https://github.com/DTU-PAS/DEAL.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。