# MegBA: A GPU-Based Distributed Library for Large-Scale Bundle Adjustment

arXiv: https://arxiv.org/abs/2112.01349 · HTML: https://mlsanity.com/paper/2112.01349/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 43 references that are in the map.
- First posted on arXiv 2 Dec 2021.

## Builds on (most cited first, 13 of 13)

- [Multi-View 3D Object Detection Network for Autonomous Driving](https://mlsanity.com/paper/1611.07759/) (2016)
- [BA-Net: Dense Bundle Adjustment Network](https://mlsanity.com/paper/1806.04807/) (2018)
- [Stereo R-CNN based 3D Object Detection for Autonomous Driving](https://mlsanity.com/paper/1902.09738/) (2019)
- [Semantic3D.net: A new Large-scale Point Cloud Classification Benchmark](https://mlsanity.com/paper/1704.03847/) (2017)
- [Accurate Monocular Object Detection via Color-Embedded 3D Reconstruction for Autonomous Driving](https://mlsanity.com/paper/1903.11444/) (2019)
- [Data-Efficient Decentralized Visual SLAM](https://mlsanity.com/paper/1710.05772/) (2017)
- [PointAR: Efficient Lighting Estimation for Mobile Augmented Reality](https://mlsanity.com/paper/2004.00006/) (2020)
- [Bundle Adjustment on a Graph Processor](https://mlsanity.com/paper/2003.03134/) (2020)
- [Stochastic Bundle Adjustment for Efficient and Scalable 3D Reconstruction](https://mlsanity.com/paper/2008.00446/) (2020)
- [CROSSBOW: Scaling Deep Learning with Small Batch Sizes on Multi-GPU Servers](https://mlsanity.com/paper/1901.02244/) (2019)
- [Square Root Bundle Adjustment for Large-Scale Reconstruction](https://mlsanity.com/paper/2103.01843/) (2021)
- [Fast and Flexible Human Pose Estimation with HyperPose](https://mlsanity.com/paper/2108.11826/) (2021)
- [RPBA -- Robust Parallel Bundle Adjustment Based on Covariance Information](https://mlsanity.com/paper/1910.08138/) (2019)

## Built on by (most cited first, 13 of 13)

- [Detector-Free Structure from Motion](https://mlsanity.com/paper/2306.15669/) (2023)
- [Power Variable Projection for Initialization-Free Large-Scale Bundle Adjustment](https://mlsanity.com/paper/2405.05079/) (2024)
- [Building Rome with Convex Optimization](https://mlsanity.com/paper/2502.04640/) (2025)
- [On-the-Fly SfM: What you capture is What you get](https://mlsanity.com/paper/2309.11883/) (2023)
- [Decentralization and Acceleration Enables Large-Scale Bundle Adjustment](https://mlsanity.com/paper/2305.07026/) (2023)
- [Distributed bundle adjustment with block-based sparse matrix compression for super large scale datasets](https://mlsanity.com/paper/2307.08383/) (2023)
- [A Taxonomy of Structure from Motion Methods](https://mlsanity.com/paper/2505.15814/) (2025)
- [Immediate 3D Gaussian Splat Reconstruction of Unordered Input with Global Consistency](https://mlsanity.com/paper/2607.14481/) (2026)
- [A Pose-only Geometric Constraint for Multi-Camera Pose Adjustment](https://mlsanity.com/paper/2604.23704/) (2026)
- [MGSfM: Multi-Camera Geometry Driven Global Structure-from-Motion](https://mlsanity.com/paper/2507.03306/) (2025)
- [CSS-BA: Gate-Guided Column Space Search for Bundle Adjustment](https://mlsanity.com/paper/2607.15652/) (2026)
- [HETA++: Global Structure-from-Motion with Hybrid Explicit Translation Averaging](https://mlsanity.com/paper/2607.15912/) (2026)
- [Accelerating Large-scale Bundle Adjustment for LiDAR Mapping via Parallel Computing](https://mlsanity.com/paper/2608.14266/) (2026)

## Abstract (arXiv)

Large-scale Bundle Adjustment (BA) requires massive memory and computation resources which are difficult to be fulfilled by existing BA libraries. In this paper, we propose MegBA, a GPU-based distributed BA library. MegBA can provide massive aggregated memory by automatically partitioning large BA problems, and assigning the solvers of sub-problems to parallel nodes. The parallel solvers adopt distributed Precondition Conjugate Gradient and distributed Schur Elimination, so that an effective solution, which can match the precision of those computed by a single node, can be efficiently computed. To accelerate BA computation, we implement end-to-end BA computation using high-performance primitives available on commodity GPUs. MegBA exposes easy-to-use APIs that are compatible with existing popular BA libraries. Experiments show that MegBA can significantly outperform state-of-the-art BA libraries: Ceres (41.45×), RootBA (64.576×) and DeepLM (6.769×) in several large-scale BA benchmarks. The code of MegBA is available at https://github.com/MegviiRobot/MegBA.