Leaderboard - LaMAria

The LaMAria test sequences can be used to evaluate the performance of SLAM algorithms in various challenges typical of egocentric data. Here, we showcase the performance rankings of different methods across the test sequences. A method appears on the summary leaderboard only when it provides results for all sequences within a challenge. For per‐sequence challenge‐wise results, select a challenge from the dropdown below. To view additional details about a specific method, click on its name in the leaderboard table.

The submissions made by the LaMAria team are marked with (r) to indicate reproduced results. Here, we submit one run out of three from our own evaluation results.

For our Aria's SLAM submission (in red indicating closed-source), we do not show pose recall due to the initialization of our ground truth pipeline with the Aria's SLAM solution.

Method	Info	Short		Medium		Long		Low light		Moving platform
Method	Info	score	R@5m	score	R@5m	score	R@5m	score	R@5m	score
GroundedSLAM	bino, imu	80.2	99.9	61.6	96.2	59.9	99.3	67.7	99.9	41.9
AnonSLAM	mono	41.9	88.1	29.8	73.9	25.5	69.5	38.4	80.3	19.2
ASTRM_SLAM_v2	mono, imu	46.0	71.3	28.8	59.6	24.8	57.4	41.4	73.1	19.8
Mighty Camera	mono, imu	35.0	66.1	27.0	53.3	18.4	41.5	34.4	70.9	11.4
ASTRM_FLOWSLAM_v1	mono, imu	46.9	76.2	30.1	56.7	26.0	58.1	34.3	59.3	23.5
ASTRM_SLAM_v0	mono, imu	43.9	70.6	40.5	72.4	26.4	59.1	29.5	43.7	23.1
(r) OpenVINS+Maplab	bino, imu	27.7	60.8	23.4	52.3	12.8	26.1	19.8	40.5	13.9
microSLAM	mono	34.4	75.4	20.3	37.6	9.1	20.0	16.8	37.2	3.8
(r) OpenVINS	bino, imu	18.9	54.2	15.9	46.8	12.5	28.1	13.9	34.9	10.4
(r) OKVIS2	bino, imu	20.0	50.0	11.6	27.9	2.6	1.4	14.5	33.0	4.7
(r) OpenVINS+Maplab	mono, imu	19.7	43.8	12.7	29.0	5.5	10.2	6.9	15.0	2.0
ViPE	mono	15.8	32.8	9.0	19.0	2.6	5.8	6.0	14.6	2.4
(r) OpenVINS	mono, imu	16.0	40.3	10.6	24.0	5.5	13.7	5.5	13.0	1.5
(r) ORB-SLAM3	mono, imu	23.0	61.2	10.9	26.0	11.2	28.7	3.1	9.0	2.0
(r) Kimera VIO	mono, imu	6.8	10.1	8.3	18.7	7.2	17.2	3.3	7.8	3.6
(r) DPV-SLAM	mono	8.0	14.1	8.8	15.6	3.9	8.7	3.0	6.8	2.3
(r) DPVO	mono	7.6	18.0	7.5	15.4	1.7	2.9	2.6	5.1	2.3
DROID-W	mono	11.4	26.1	2.7	5.8	0.0	0.0	0.4	1.2	1.6
LingBot-Map	mono	1.6	2.4	0.7	0.9	0.3	0.3	1.2	1.2	0.4
cusvio	bino, imu	42.7	81.0	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
Test 4	mono	7.9	3.0	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
MVIO	bino, imu	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
(r) Aria's SLAM	bino, imu	90.7	N.A.	78.5	N.A.	70.9	N.A.	84.2	N.A.	55.0

Changelog

Updates regarding leaderboard maintenance, evaluation changes, and data releases.

2026-10-09 — Throughput (frames per second) is now a compulsory parameter in the submission form.
2026-10-07 — Fix inflated pose recall for incomplete trajectories. Recall was inflated when an estimated trajectory was shorter than the ground truth, all ground-truth keyframe poses count in recall. Complete trajectories and control point / ATE scores are unchanged.
2026-09-15 — Bugfix: Control Point Triangulation. Fixes an evaluation issue introduced by a refactoring. Original leaderboard runs used the original code and are unaffected.