Where Grounding Accuracy Lives on the IoU Curve: Label-Free Inference-Time Boundary Refinement

A new arXiv paper (2608.19553v1) introduces label-free precision refinement (LFPR), a method that improves grounding bounding-box accuracy by routing predicted-small regions to a higher-resolution pass without accessing target annotations at inference. On 31,921 Ref-L4 expressions, LFPR raises mAcc0.5:0.95 from 72.947% to 76.013%, and on 30,969 RefCOCO/RefCOCO+/RefCOCOg expressions it improves every dataset at Acc@0.5, mAcc, and mean IoU, while a prospective Flickr30K Entities evaluation improves every endpoint (mAcc +0.973, Acc@0.9 +1.022). The authors show that referent selection and boundary precision are partially separable, with different components moving opposing regions of the IoU curve.

arXiv:2608.19553v1 Announce Type: new Abstract: Vision--language models can identify the correct referent while returning an imprecise bounding box. We study whether a frozen direct-answer model can use its own prediction to allocate one additional localized observation without accessing target annotations at inference. Label-free precision refinement LFPR routes predicted-small regions to a higher-resolution pass, re-grounds the expression inside a context crop, admits a candidate only under fixed geometric guards, and returns a fixed coordinate-wise midpoint. We report results across three evidence tiers. On 31,921 retrospective Ref-L4 expressions, LFPR raises mAcc$ {0.5:0.95}$ from 72.947\% to 76.013\% Acc@0.5 88.531\%$\to$89.725\%, Acc@0.9 55.788\%$\to$61.142\% . A frozen transfer to 30,969 RefCOCO/RefCOCO+/RefCOCOg expressions improves every dataset at Acc@0.5, mAcc, and mean IoU pooled mAcc $+0.645$, Acc@0.5 $+0.817$ , while Acc@0.9 is unchanged overall: routing alone gains $+1.162$ points there, but crop, guards, and fusion give back $-1.192$, offsetting rather than showing no strict-IoU effect. A prospective, image-disjoint Flickr30K Entities evaluation improves every endpoint mAcc $+0.973$, Acc@0.9 $+1.022$ , more strongly under a single-box variant mAcc $+2.575$, Acc@0.9 $+3.689$ . The same operator applied to two released grounding specialists improves every endpoint Acc@0.9 $+1.569$/$+6.716$ for EGM-4B/8B at roughly twice the latency, composing with specialist training rather than replacing it. A genuine unguarded control guard removed from the same candidates underperforms the incumbent on every metric, showing the guard is load-bearing. Together, these results show that referent selection and boundary precision are partially separable, with different components moving opposing regions of the IoU curve -- behavior a single threshold cannot reveal.