Method Detail: VLMod-Obstacle-2026

Back to Leaderboard
Benchmark: VLMOD
Short name: VLMod-Obstacle-2026
Long name: VLMod-3D-Obstacle-Tracker
Description: a lightweight framework for multi-camera spatial obstacle detection and track annotation across sequential video streams. Designed to process high-resolution multi-view camera feeds, the method filters static frame padding and extracts sparse non-zero detection vectors. By mapping timestamped bounding data (Dataset_Camera_Res_StartTS_EndTS_ObstacleID) across distributed camera setups, VLMod enables rapid spatial localization and high-density tracking for autonomous perception systems.
Reference: VLMod Development Team, VLMod: Multi-Camera Spatial Obstacle Detection and Vector Tracking. In arXiv preprint arXiv:2609.00000, 2026.
Last submitted: September 21, 2026
Published: September 16, 2026 at 06:56:33
Submissions: 8
Project page / code: https://github.com/zainab1kausar338-babyseven/VLMod-3D-Obstacle-Tracker
Open source: Yes

Benchmark performance

Submission Date F1 (↑) Precision (↑) Recall (↑) TP (↑) FP (↓) FN (↓)
2026-09-21 18:26 - - - - - -
2026-09-18 08:09 - - - - - -
2026-09-18 07:38 - - - - - -
2026-09-18 07:31 - - - - - -
2026-09-16 09:40 26.8251 39.6805 20.2610 621 944 2444
2026-09-16 09:21 33.7208 29.5388 39.2822 1204 2872 1861
2026-09-16 07:23 39.1178 31.1163 52.6591 1614 3573 1451
2026-09-16 07:07 11.7810 38.6570 6.9494 213 338 2852