Dataset Overview
Photo Modeling DatasetYes.Changsha Langhui Information Technology Co., Ltd. DataAssetsAPI platform operates. High-precision 3D models covering 1,200+ scenarios and their corresponding 7.25+ million multiple-viewed high-overlay sequence photographs (approximately 28TB) are applicable to 3D reconstruction of the landscape, digitization of cultural objects and reverse modelling of buildings.
| Dataset Name | Photo Modeling Dataset |
| Number of 3D models | 1200+ (approximately 3950GB) |
| Number of photographs | 7.25+10,000 (approximately 2528GB) |
| Data Type | Image (multi-perspective high-overlapping sequence photographs) + 3D model |
| Data Source | I'm taking the scenery. |
| Data Status | Fully Cleaned |
| Apply Tabs | Field adapted language |
Core Data Fields
Each piece of data contains complete structured fields that cover everything from raw data to annotated information.
| Field Category | Field Count | Typical Fields |
|---|---|---|
| Sequence Photo | 8 | Multi-perspective sequence images, overlap (>60%), camera parameters (focal/sensor size), range of the photograph |
| 3D model | 10 | Model file (OBJ/FBX/PLY, etc.), face number/cap number, texture paste, coordinate system, accuracy level |
| ScenesMetadata | 8 | Type of scene (building/antiquities/natural landscape/industrial facilities), environmental lighting conditions, date of shooting |
| Geometric Information | 6 | Ground sampling distance (GSD), cloud density, area covered, model scale |
Data Sample Showcase
The following is a structured JSON sample showing typical data formats.
Other Organiser
"scene id": "PM-2024-000123",
"scene type": "An ancient architecture," "scene name": "A provincial temple of temples,"
"Photos": { "count": 3840, "avg overlap pct": 72, "gsd mm": 2.3, "total size gb": 13},
"Model": "format": "OBJ+JPG", "vertices": 24.5 million, "faces": 487 million, "texture resolution": "8192x8192", "accuracy level": "high precision",
"coverage": {"scene area m2": 450, "model scale": "1:1"}
♪ I'm sorry ♪AI Application Scenarios
3D modelling of the scenery
Massive multi-view photo training and optimization of the Structure from Motion and multi-view stereo-matching algorithm.
Digital protection of cultural objects
High-precision models of ancient buildings/antiquities are used for the digital archiving and virtual display of cultural heritage 3D.
Building reverse modelling
The building scene photo and model pairing training AAI automolemic algorithm.
Optimization of tilted photo modelling
Multi-perspective data are used for training and precision enhancement of three-dimensional reconstruction algorithms for tilt photography.
Small scenes are fine reconstruction
Small scene data training refines 3D reconstruction models.
Digital twin cities
The external architecture model is the basic data for digital twin cities.
Frequently asked questions
Build your digital twin world?
Langham Technology has the nation's leading real-world 3D data acquisition capabilities, collecting a total of 2000 + square kilometers of flight data and 1200 + fine models. All data has been cleaned and quality inspected and can be used directly for AI model training.
Data sample preview · Customized acquisition plan · Professional technical support