Skip to content
RobotWorld
← Back to Videos

Boxer Lifts 2D Detections into Fused 3D Boxes

Loading video
Loading video

Meta Reality Labs' Boxer lifts open-vocabulary 2D detections into globally fused 3D oriented boxes from posed images and point clouds, with offline fusion and online tracking.

3D Object DetectionOpen-Vocabulary PerceptionOriented Bounding BoxesEgocentric VisionMeta Reality LabsBoxer
Category: perception
Author: @ddetone
Date: 2026-09-12T00:00:00
Duration: 16.66s
Source: https://x.com/ddetone/status/2098677887108776256
Reference: https://facebookresearch.github.io/boxer/

Related Videos

EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience
2:40

EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience

@liuziwei7 · 108 views · 2026-10-20
Egocentric Vision3D reconstruction
MINT open-source egocentric hand motion reconstruction
1:22

MINT open-source egocentric hand motion reconstruction

@wuji_global · 583 views · 2026-09-10
Motion CaptureEgocentric Vision
Hesai Kosmo: LiDAR + 8K Capture for Cloud 3D Reconstruction
0:36

Hesai Kosmo: LiDAR + 8K Capture for Cloud 3D Reconstruction

@tokufxug · 64 views · 2026-10-25
3D reconstruction3D Gaussian Splatting
Catching LiDAR Ghost Cars by Their Missing Shadow
0:33

Catching LiDAR Ghost Cars by Their Missing Shadow

@AlloyRobotics · 52 views · 2026-10-24
LiDARAdversarial Attack
SpatialScout Benchmarks Active 3D Exploration for VLM Agents
0:30

SpatialScout Benchmarks Active 3D Exploration for VLM Agents

@HuggingPapers · 88 views · 2026-10-22
Embodied AIVLM
OpenTac Opens the Full Stack for 3D Tactile Sensing
1:05

OpenTac Opens the Full Stack for 3D Tactile Sensing

@arikslep · 102 views · 2026-10-21
Tactile SensingElectronic skin

As an Amazon Associate, we earn from qualifying purchases.