无需训练的图像分割框架，自动确定聚类数并优化边界
CLASP: Adaptive Spectral Clustering for Unsupervised Per-Image Segmentation
- 用自监督ViT提取像素块特征，结合谱聚类进行无监督分割
- 在COCO Stuff和ADE20K上达到与最新无监督方法相当的性能
- 无需调参且零训练，适合广告、内容审核等实际场景
我们提出CLASP（基于自适应谱处理的聚类），一种轻量级的无监督图像分割框架，不依赖任何标注数据或微调。CLASP首先使用自监督ViT编码器（DINO）提取每个图像块的特征；然后构建亲和矩阵并应用谱聚类。为避免手动调参，采用特征间隙轮廓搜索法自动确定聚类数量，并通过全连接的DenseCRF锐化分割边界。尽管结构简单且无需训练，CLASP在COCO Stuff和ADE20K数据集上仍达到了具有竞争力的平均交并比（mIoU）和像素准确率，与近期无监督基线方法相当。其零训练设计使其成为大规模未标注数据集的强基准，特别适用于数字广告与营销工作流中的品牌安全筛查、创意素材筛选及社交媒体内容审核。
原文摘要 · Abstract (English)
We introduce CLASP (Clustering via Adaptive Spectral Processing), a lightweight framework for unsupervised image segmentation that operates without any labeled data or finetuning. CLASP first extracts per patch features using a self supervised ViT encoder (DINO); then, it builds an affinity matrix and applies spectral clustering. To avoid manual tuning, we select the segment count automatically with a eigengap silhouette search, and we sharpen the boundaries with a fully connected DenseCRF. Despite its simplicity and training free nature, CLASP attains competitive mIoU and pixel accuracy on COCO Stuff and ADE20K, matching recent unsupervised baselines. The zero training design makes CLASP a strong, easily reproducible baseline for large unannotated corpora especially common in digital advertising and marketing workflows such as brand safety screening, creative asset curation, and social media content moderation
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。