M:3L Lab

ICRA 2026

TUN3D: Towards Real-World Scene Understanding from Unposed Images

The first method for joint layout estimation and 3D detection without camera poses or depth supervision, designed to work with point clouds, posed images, or unposed RGB captures.

Publication record

Status
Accepted at ICRA 2026
Authors
Anton Konushin, Nikita Drozdov, Bulat Gabdullin, Alexey Zakharov, Anna Vorontsova, Danila Rukhovich, Maksim Kolodiazhnyi
Venue
ICRA 2026
Research area
3D Intelligence
First arXiv submission

Abstract / summary

TUN3D jointly estimates indoor scene layouts and detects 3D objects from point clouds, posed images, or fully unposed multi-view RGB. In the unposed setting it does not require known camera poses, camera calibration, or externally provided depth as input.

Methodology

  • Lightweight sparse-convolutional scene backbone
  • Dedicated 3D object-detection and wall-layout heads
  • A bird's-eye-view parametric wall representation with two 2D corner offsets and wall height
  • DUSt3R reconstruction for posed or fully unposed image collections
  • Joint optimization of object classification, 3D boxes, wall classification, and wall parameters

Datasets and benchmarks

  • ScanNet
  • S3DIS
  • ARKitScenes
  • Structured3D

Principal results

  • The paper presents the first reported joint layout-estimation and 3D-detection results on real scans from fully unposed images.
  • On unposed ScanNet imagery, TUN3D reports layout F1 of 46.5, object-detection [email protected] of 44.0, and [email protected] of 20.7.
  • With ground-truth point clouds, the paper reports state-of-the-art layout results across ScanNet, S3DIS, ARKitScenes, and Structured3D at publication time.

Benchmark evidence

BenchmarkTaskInputMetricResultEvidence
ScanNetJoint layout estimation and 3D object detectionUnposed multi-view RGB; no known poses or external depthLayout F1 / detection [email protected] / [email protected]46.5 / 44.0 / 20.7TUN3D paper, Table I
S3DISJoint layout estimation and 3D object detectionUnposed multi-view RGB; no known poses or external depthLayout F1 / detection [email protected] / [email protected]20.8 / 11.0 / 2.2TUN3D paper, Table I
ScanNetJoint layout estimation and 3D object detectionPosed multi-view RGB; no ground-truth depthLayout F1 / detection [email protected] / [email protected]55.2 / 57.4 / 35.6TUN3D paper, Table I
ARKitScenesLayout estimationGround-truth point cloud; cross-dataset evaluationLayout F130.3TUN3D paper, Table II
The paper reports 25.9 for Omni-PQ under the same table protocol.
Structured3DLayout estimation and 3D object detectionGround-truth point cloudLayout [email protected]/[email protected]; detection [email protected]/[email protected]90.5/89.6; 73.9/65.4TUN3D paper, Table III

Values are attributed to the associated paper and the stated input and evaluation protocol.

Technologies and methods

  • Sparse convolutional networks
  • DUSt3R
  • TSDF fusion
  • Finite-dimensional parametric wall models

Related M:3L research

Evidence boundary

Benchmark claims and numerical values are paper-reported results for the named protocol. Potential applications are not evidence of completed deployments.