PCA-Seg：重新审视开放词表语义分割与部件分割中的代价聚合
PCA-Seg: Revisiting Cost Aggregation for Open-Vocabulary Semantic and Part Segmentation
把开放词表分割中「空间聚合→类别聚合」的串行代价体处理改为并行双流（语义流+上下文流）加正交化解耦，并用系数映射器做像素级自适应加权；做开放词表分割或需要从图文代价体中抽取对齐线索的工作可直接复用该并行聚合块（仅增 0.35M 参数）。
原文依据 existing methods extract image-text alignment cues from cost volumes through a serial structure of spatial and class aggregations
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期视觉语言模型（VLM）的进展在开放词表语义分割与部件分割（OSPS）领域引起了广泛关注。然而，现有方法通过空间聚合与类别聚合的串行结构从代价体中提取图文对齐线索，导致类别级语义与空间上下文之间产生知识干扰。为此，本文提出一种简单而有效的并行代价聚合（PCA-Seg）范式以缓解上述挑战，使模型能够从代价体中捕获更丰富的视觉语言对齐信息。具体而言，我们设计了一个专家驱动的感知学习（EPL）模块，可高效整合语义流与上下文流。该模块引入多专家解析器，从多个视角提取互补特征。此外，我们设计了系数映射器，以自适应地为每个特征学习像素特定的权重，从而将互补知识整合为统一且鲁棒的特征嵌入。进一步地，我们提出特征正交化解耦（FOD）策略，以减轻语义流与上下文流之间的冗余，使 EPL 模块能够从正交化特征中学习多样化的知识。在八个基准上的大量实验表明，PCA-Seg 中的每个并行块仅增加 0.35M 参数，即达到了最优的 OSPS 性能。
英文原文摘要 1427 字符
Recent advances in vision-language models (VLMs) have garnered substantial attention in open-vocabulary semantic and part segmentation (OSPS). However, existing methods extract image-text alignment cues from cost volumes through a serial structure of spatial and class aggregations, leading to knowledge interference between class-level semantics and spatial context. Therefore, this paper proposes a simple yet effective parallel cost aggregation (PCA-Seg) paradigm to alleviate the above challenge, enabling the model to capture richer vision-language alignment information from cost volumes. Specifically, we design an expert-driven perceptual learning (EPL) module that efficiently integrates semantic and contextual streams. It incorporates a multi-expert parser to extract complementary features from multiple perspectives. In addition, a coefficient mapper is designed to adaptively learn pixel-specific weights for each feature, enabling the integration of complementary knowledge into a unified and robust feature embedding. Furthermore, we propose a feature orthogonalization decoupling (FOD) strategy to mitigate redundancy between the semantic and contextual streams, which allows the EPL module to learn diverse knowledge from orthogonalized features. Extensive experiments on eight benchmarks show that each parallel block in PCA-Seg adds merely 0.35M parameters while achieving state-of-the-art OSPS performance.