CLON: Cue-Calibrated Linguistic Object Onboarding for Zero-Shot 6D Pose Front-Ends

SummaryAI generated

On September 7, 2026, arXiv cs.CV published the paper CLON (Cue-Calibrated Linguistic Object Onboarding for Zero-Shot 6D Pose Front-Ends). This method aims to solve the front-end problem of 6D pose estimation in zero-sample scenarios by introducing linguistic cues for object calibration, thereby enhancing the model’s generalization ability on unseen objects or environments.

Related eventsRELATED EVENTS

2026-09-09 05:29OpenAI claims huge maths breakthrough on a famed ‘Millennium Problem’
2026-09-09 06:11OpenAI claims to have solved the Navier-Stokes equation problem, but its originality is questioned by the academic community
2026-09-08 22:00Google DeepMind Maps 9 Billion Possible DNA Variants

Key entitiesKEY ENTITIES

BOP-Classic-CoreCLONSAM 3

Coverage · reports per dayLANGUAGE SPLIT

Entity relations

SignalsSIGNALS

Keyword heat

CLON1
SAM 31
BOP-Classic-Core1

All reports (1)SOURCES

↗

A arXiv cs.CV en 2026-09-07 12:00

CLON 提出一种无需任务特定训练即可为新物体上载的前端模块，在七个 BOP-Classic-Core 数据集上使检测 AP 提升 8.1 个百分点、分割 AP 提升 6.2 个百分点，下游 6D 姿态 AR 最高提升 4.1 个百分点。CLON 利用渲染模板构建语言语义记忆以引导 SAM 3 生成高召回率提议，并计算一次固定物体集提示权重进行校准评分。该方法无需针对新物体训练，旨在保留部分可见真阳性同时拒绝语义合理干扰项，其性能优于 CNOS 和 SAM-6D 前端。