arXiv：cs.AI· Zixun Wang, Ben Dai·· 5 小时前
RankSEG 用空间局部依赖结构替代条件独立假设，改进图像分割
On the Relaxation of Conditional Independence Assumption for Image Segmentation
AI 导读
针对 RankSEG 依赖的条件独立假设（CIA）忽略标签相关性、在低对比度场景下性能下降的问题，研究者提出空间局部依赖（SLD）结构，并配合倒数矩近似与不动点优化策略，将复杂度降至 O(d log d)。该算法在多个分割基准上稳定优于 argmax 和基于 CIA 的 RankSEG，在低对比度或小目标场景中提升显著。
正文
Abstract:In semantic segmentation, a recent line of RankSEG methods directly optimizes Dice/IoU scores at inference time, improving alignment with evaluation metrics without modifying model training. Despite its theoretical and empirical success, RankSEG relies on the restrictive Conditional Independence Assumption (CIA), which ignores crucial label correlations and therefore degrades performance in ambiguous or low-contrast scenarios. However, accounting for full label dependence is computationally prohibitive, requiring $\mathcal{O}(d^3)$ time. To address this, we replace the CIA with a Spatially Localized Dependence (SLD) structure that captures local label correlations while keeping the dependence model tractable. We further overcome the remaining computational bottleneck via a Reciprocal Moment Approximation coupled with a novel fixed-point optimization strategy that eliminates exhaustive search. The proposed algorithm achieves a highly practical $\mathcal{O}(d \log d)$ complexity and consistently outperforms conventional argmax and CIA-based RankSEG across diverse segmentation benchmarks. Improvements are significant in low-contrast or small-object scenarios, where label dependence offers valuable signals complementary to image information for accurate segmentation. The code of experiments is available at this https URL.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML) |
| Cite as: | arXiv:2609.38930 [cs.CV] |
| (or arXiv:2609.38930v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2609.38930 arXiv-issued DOI via DataCite |
Submission history
From: Zixun Wang [view email]
[v1]
Wed, 30 Sep 2026 04:10:57 UTC (34,382 KB)
[v2]
Mon, 12 Oct 2026 06:52:29 UTC (34,390 KB)
来源：arXiv：cs.AI · arxiv.org