Papers/2609.03102
🧪 Test?View on arXiv

WireSeg-32K: A Physics-Grounded Synthetic Dataset for Wire Instance Segmentation

Not provided

synthetic datainstance segmentationfine-tuning
2609.03102
Builder Relevance
80%
1h ago

Abstract

WireSeg-32K is a synthetic dataset designed to improve wire instance segmentation by providing 32,000 RGB images, instance masks, depth maps, and a real-world test set with annotations.

Reality Card

Core Claim

The introduction of WireSeg-32k significantly improves wire instance segmentation performance, achieving a 10.2% increase in real-world mAP@75 through the fine-tuning of SAM3.

Method / Result

Fine-tuning SAM3 on WireSeg-32k improved real-world mAP@75 by 10.2%.

Limitations

The dataset relies on a synthetic generation process, which may not fully capture the complexities of real-world wire scenarios.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers