Notice: This website is continuously updated as new conference proceedings and journal volumes are published. Consequently, the figures presented here may differ from those reported in our publications.
A brief overview of our methodology is provided on the Methods page, along with references to publications documenting our approach and links to the code and data used in each study.
The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates.
Cycle-Sync: Robust Global Camera Pose Estimation through Enhanced Cycle-Consistent Synchronization
Authors: Shaohan Li, Yunpeng Shi, Gilad Lerman
NeurIPS 2025 | Venue PDF | LLM Run Details | Input Tokens: 30,678 Total number of tokens sent to the LLM as input for this paper's analysis. | Output Tokens: 4,115 Total number of tokens produced by the LLM (including reasoning/thinking tokens) for this paper's analysis.
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on synthetic and real datasets show that Cycle-Sync consistently outperforms leading pose estimators, including full structure-frommotion pipelines with bundle adjustment. |
| Researcher Affiliation | Academia | Shaohan Li* Yunpeng Shi Gilad Lerman* *School of Mathematics, University of Minnesota Department of Mathematics, University of California, Davis EMAIL, EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Cycle-Sync Input: {γij}ij E, { dij,k}k Cij, β, {λt}t 1, δ (default: 10 8) Steps: Compute {sij}ij E by T-AAB While t tmax: t = t + 1 ... |
| Open Source Code | Yes | 1The Matlab code is released at https://github.com/sli743/Cycle-Sync |
| Open Datasets | Yes | We conduct real-world experiments on 13 ETH3D stereo datasets [23, 24]... We also compare the camera location estimation results for different location estimation algorithms on IMC-PT. |
| Dataset Splits | No | The paper uses synthetic data generated with parameters n, q, σ, p, and evaluates on 13 ETH3D stereo datasets and the IMC-PT dataset, but does not specify explicit training/testing/validation splits in terms of percentages or sample counts for its own methodology. |
| Hardware Specification | Yes | We conduct real-world experiments on 13 ETH3D stereo datasets [23, 24], using a personal laptop equipped with an 11th Gen Intel(R) Core(TM) i9-11900H processor (2.50 GHz, 8 cores, 16 threads) and 16 GB physical memory. |
| Software Dependencies | No | The paper mentions software like Matlab code, SIFT, RANSAC, COLMAP, and LoFTR [29] but does not provide specific version numbers for any of these components. |
| Experiment Setup | Yes | In all of our experiments, we choose tmax = 20, β = 20 and λt = t/(t + 10). |