BRIDGE -- Building Reinforcement-Learning Depth-to-Image Data Generation Engine for Monocular Depth Estimation

2025/09/29 by Dingning Liu, Liu, Dingning, Haoyu Guo +5
Computer Science · Engineering · #Advanced Vision and Imaging #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Processing Techniques and Applications #Industrial Vision Systems and Defect Detection

paper · pdf · doi:10.48550/arxiv.2509.25077

openalex publication_date 2025/09/29 · openalex created_date 2025/10/19 · openalex updated_date 2026/07/28

Abstract

Monocular Depth Estimation (MDE) is a foundational task for computer vision. Traditional methods are limited by data scarcity and quality, hindering their robustness. To overcome this, we propose BRIDGE, an RL-optimized depth-to-image (D2I) generation framework that synthesizes over 20M realistic and geometrically accurate RGB images, each intrinsically paired with its ground truth depth, from diverse source depth maps. Then we train our depth estimation model on this dataset, employing a hybrid supervision strategy that integrates teacher pseudo-labels with ground truth depth for comprehensive and robust training. This innovative data generation and training paradigm enables BRIDGE to achieve breakthroughs in scale and domain diversity, consistently outperforming existing state-of-the-art approaches quantitatively and in complex scene detail capture, thereby fostering general and robust depth features. Code and models are available at https://dingning-liu.github.io/bridge.github.io/.

Citations

• Distill Any Depth: Distillation Creates a Stronger Monocular Depth Estimator
• MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision
• Depth Any Video with Scalable Synthetic Data
• Depth Pro: Sharp Monocular Metric Depth in Less Than a Second
• Video Diffusion Alignment via Reward Gradients
• Depth Anything V2
• GauStudio: A Modular Framework for 3D Gaussian Splatting and Beyond
• UniDepth: Universal Monocular Metric Depth Estimation
• GeoWizard: Unleashing the Diffusion Priors for 3D Geometry Estimation from a Single Image
• Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data
• pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction
• Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation
• IEBins: Iterative Elastic Bins for Monocular Depth Estimation
• NDDepth: Normal-Distance Assisted Monocular Depth Estimation
• GEDepth: Ground Embedding for Monocular Depth Estimation
• Shrinking Class Space for Enhanced Certainty in Semi-Supervised Learning
• Learning Depth Estimation for Transparent and Mirror Surfaces
• 3D Gaussian Splatting for Real-Time Radiance Field Rendering
• MiDaS v3.1 -- A Model Zoo for Robust Monocular Relative Depth Estimation
• High-Resolution Synthetic RGB-D Datasets for Monocular Depth Estimation
• DINOv2: Learning Robust Visual Features without Supervision
• DDP: Diffusion Model for Dense Visual Prediction
• Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo
• ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth
• Adding Conditional Control to Text-to-Image Diffusion Models
• Adding Conditional Control to Text-to-Image Diffusion Models
• The Relative Importance of Depth Cues and Semantic Edges for Indoor Mobility Using Simulated Prosthetic Vision in Immersive Virtual Reality
• P3Depth: Monocular Depth Estimation with a Piecewise Planarity Prior
• Swin Transformer V2: Scaling Up Capacity and Resolution
• Vision Transformers for Dense Prediction
• Improved Denoising Diffusion Probabilistic Models
• Learning to Recover 3D Scene Shape from a Single Image
• Hypersim: A Photorealistic Synthetic Dataset for Holistic Indoor Scene Understanding
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• Denoising Diffusion Implicit Models
• Understanding SSIM
• Denoising Diffusion Probabilistic Models
• Virtual KITTI 2
• FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
• BlendedMVS: A Large-scale Dataset for Generalized Multi-view Stereo Networks
• Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset Transfer
• Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-Shot Cross-Dataset Transfer
• Unsupervised Data Augmentation for Consistency Training
• FastDepth: Fast Monocular Depth Estimation on Embedded Systems
• Pseudo-LiDAR from Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Driving
• Deep Ordinal Regression Network for Monocular Depth Estimation
• Image Matching Using SIFT, SURF, BRIEF and ORB: Performance Comparison for Distorted Images
• ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes
• Semi-Supervised Deep Learning for Monocular Depth Map Prediction
• Unsupervised Monocular Depth Estimation with Left-Right Consistency
• Single-Image Depth Perception in the Wild
• Predicting Depth, Surface Normals and Semantic Labels with a Common\n Multi-Scale Convolutional Architecture
• Unsupervised Domain Adaptation by Backpropagation
• Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
• Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
• Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

Related

• StarryGazer: Leveraging Monocular Depth Estimation Models for Domain-Agnostic Single Depth Image Completion
• Monocular Depth Estimation Using Cues Inspired by Biological Vision Systems
• BridgeNet: A Joint Learning Network of Depth Map Super-Resolution and Monocular Depth Estimation
• Refine and Distill: Exploiting Cycle-Inconsistency and Knowledge\n Distillation for Unsupervised Monocular Depth Estimation
• To complete or to estimate, that is the question: A Multi-Task Approach to Depth Completion and Monocular Depth Estimation