Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• OVI-MAP decouples instance reconstruction from semantic inference to enable stable instance tracking and zero-shot semantic labeling.method 95%
• OVI-MAP incrementally constructs a class-agnostic 3D instance map from RGB-D input.method 95%
• Semantic features in OVI-MAP are extracted only from a small set of automatically selected views using vision-language models.method 95%
• OVI-MAP operates in real time.technical 90%
• OVI-MAP outperforms state-of-the-art open-vocabulary mapping baselines on standard benchmarks.result 95%
• Existing methods often rely on the closed-set assumption or dense per-pixel language fusion, limiting scalability and temporal consistency.limitation 90%
• OVI-MAP achieves comparable or better instance segmentation accuracy than offline approaches while consistently outperforming prior online systems, particularly at high IoU thresholds (AP75).result 95%
• OVI-MAP extracts semantic features from only a small set of automatically selected views using vision-language models.method 95%
• OVI-MAP decouples instance reconstruction from semantic inference to enable stable instance tracking and zero-shot semantic labeling.method 95%
• OVI-MAP operates in real time.technical 95%
• OVI-MAP outperforms state-of-the-art open-vocabulary mapping baselines on standard benchmarks.result 95%
• OVI-MAP achieves comparable or better instance segmentation accuracy than offline approaches on Replica and ScanNet datasets, while consistently outperforming prior online systems.result 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card