# Image Alignment on APAP-railtracks (train/test/overlapping) Source: [Image Alignment on APAP-railtracks (train/test/overlapping)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:00:16.334Z Evaluation of Image Alignment performance on the APAP-railtracks dataset, covering training, testing, and overlapping subsets, typically measured using RMSE and outlier percentage. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Image Alignment | | Dataset (including any stated split) | APAP-railtracks (train/test/overlapping) | | Metric | RMSE (TR) | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 6 of 6 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [RMSE (TR)](): Lower is better; unit: SCALAR. - [RMSE (TE)](): Lower is better; unit: SCALAR. - [Outlier %](): Lower is better; unit: PERCENT. ## Recorded results | Method | RMSE (TR) | Result date | Report paper | | --- | --- | --- | --- | | Mesh-based Warp | 3.23 | Not recorded | 68fa66581e209131cea0faae | | Mesh-based Warp | 3.23 | Not recorded | 68fa66581e209131cea0faae | | APAP | 5.43 | Not recorded | 68fa66581e209131cea0faae | | APAP | 5.43 | Not recorded | 68fa66581e209131cea0faae | | Homography | 5.58 | Not recorded | 68fa66581e209131cea0faae | | Homography | 5.58 | Not recorded | 68fa66581e209131cea0faae | ### Mesh-based Warp RMSE (TR): 3.23. Result ID: 698ceb2d8b517eb4331dac7e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"dual-feature":false,"grid size":"40x40"}. This record does not assert independent reproduction. ### Mesh-based Warp RMSE (TR): 3.23. Result ID: 698ceb2d8b517eb4331dac7f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"dual-feature":true,"grid size":"40x40"}. This record does not assert independent reproduction. ### APAP RMSE (TR): 5.43. Result ID: 698ceb2d8b517eb4331dac7c. [Reporting paper]() Recorded evaluation configuration: {"dual-feature":false}. This record does not assert independent reproduction. ### APAP RMSE (TR): 5.43. Result ID: 698ceb2d8b517eb4331dac7d. [Reporting paper]() Recorded evaluation configuration: {"dual-feature":true}. This record does not assert independent reproduction. ### Homography RMSE (TR): 5.58. Result ID: 698ceb2d8b517eb4331dac7a. [Reporting paper]() Recorded evaluation configuration: {"dual-feature":false}. This record does not assert independent reproduction. ### Homography RMSE (TR): 5.58. Result ID: 698ceb2d8b517eb4331dac7b. [Reporting paper]() Recorded evaluation configuration: {"dual-feature":true}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.