cs.CVNov 13, 2026

Signal Scaffolding: Unlocking the Potential of Lightweight Models for Self-Supervised Monocular Depth Estimation

Authors: Zhijie Wang, Li Zhang, Yujie Chen, Shibiao Deng, Xin Gao, Houzhi Zhou, Jiangshuo Zhang

Organizations: Department of Mathematics, Hefei University of Technology Hefei, China

Abstract

With the rising demand for edge devices, lightweight self-supervised monocular depth estimation methods have continued to evolve; however, existing methods generally suffer from poor accuracy. We find that the root cause may lies in the failure of conventional training regimes to fully exploit the inherent potential of lightweight models. To address this, we propose a novel three-phase training strategy for lightweight self-supervised monocular depth estimation, termed Signal Scaffolding (SS). Without introducing any additional dataset or parameters, we apply SS to ReForgeDepth, a lightweight network of our own design. Experiments on the KITTI and Cityscapes datasets demonstrate that our method achieves state-of-the-art performance in the accuracy-efficiency trade-off and running in real time on edge devices. These results suggest that lightweight models can be improved further within their existing capacity, and that the central question may be less about how parameters are allocated than about how to unlock this latent potential through training.

Figures & tables

Explore similar work

CardsList
  1. XiDepth: a Lightweight and Efficient Network for Self-supervised Monocular Depth Estimation

    Aug 4, 2026Elena Izzo, Riccardo Toniolo, Lamberto BallanEnergy-Efficient MLMonocular Depth Estimation

  2. ZipDepth: Bringing Lightweight Zero-Shot Monocular Depth Anywhere, on Any Device

    Jul 9, 2026Fabio Tosi, Luca Bartolomei, Matteo Poggi +1Efficient Neural Network InferenceMonocular Depth Estimation

  3. CMAMBADEPTH: Self-supervised Monocular Depth Estimation with Channel Mamba and Hybrid Attention

    Sep 21, 2026Xuezhi Xiang, Jiayao Liu, Heqi Xiang +3Monocular Depth EstimationMulti-Scale Feature Fusion