GIFT: Geometry-Invariant Fine-Tuning for Non-Lambertian Monocular Depth Estimation
2026-08-03 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors noticed that monocular depth models often make mistakes on shiny or see-through surfaces like mirrors and glass because they predict depth based on what is reflected or seen through instead of the actual surface. To fix this, they created a method called GIFT that uses multiple photos taken with the same camera and setup but with different appearances, so the model learns the true shape that doesn't change. Importantly, GIFT doesn't need precise depth measurements to improve these models. Their tests show that GIFT reduces errors on tricky surfaces while keeping good performance elsewhere.
Monocular depth estimationNon-Lambertian surfacesDepth hallucinationGeometry invarianceFine-tuningSynthetic training dataDepth sensorsMirrors and glassParameter-efficient adaptationBenchmark evaluation
Authors
Xianghui Fan, Zhaoyu Chen, Bingqian Wu, Dayu Li, Xin Zeng, Huanran Cui, Guangzhen Xu, Xiangru Huang, Hang Yang
Abstract
Monocular depth foundation models, benefiting from large-scale synthetic training data, have demonstrated strong generalization. However, they often hallucinate depth on non-Lambertian surfaces, estimating reflected content in mirrors or transmitted content behind glass rather than the physical surface itself. Adapting these models with real-world data is challenging because conventional depth sensors are also unreliable in such regions. We observe that while the appearance of a non-Lambertian surface varies with its reflected or transmitted environment, its underlying geometry remains unchanged. Based on this observation, we propose GIFT (Geometry-Invariant Fine-Tuning), a parameter-efficient post-training framework that requires no measured depth labels. We collect groups of RGB images under controlled appearance changes while keeping the camera and target geometry fixed. GIFT exploits geometric invariance across these observations to suppress non-Lambertian depth hallucinations while retaining general depth estimation capability. We further construct a controlled benchmark that evaluates non-Lambertian depth recovery, robustness to appearance changes, and performance retention in other regions. Experiments on our benchmark and an independent real-world dataset demonstrate that GIFT improves depth prediction for mirrors and transparent objects while largely preserving the base model's performance, providing a practical and low-cost approach for adapting monocular depth foundation models to non-Lambertian scenes.