多层特征融合提升无监督物体分割精度
MUFASA: A Multi-Layer Framework for Slot Attention
- 从ViT多层特征中提取槽注意力，挖掘深层语义信息
- 在多个数据集上实现更优分割效果，收敛更快
- 轻量级插件设计，推理开销极小
无监督对象中心学习（OCL）将视觉场景分解为独立实体。槽注意力是一种流行方法，通过称为“槽”的潜在向量表示单个对象。现有方法仅从预训练视觉变换器（ViT）的最后一层获取这些槽表示，忽略了其他层中蕴含的丰富语义信息。为更好利用这些潜在语义信息，我们提出MUFASA——一种面向基于槽注意力的无监督对象分割的轻量级即插即用框架。该模型在ViT编码器的多个特征层上计算槽注意力，充分挖掘其语义丰富性，并提出一种融合策略，将多层获得的槽聚合为统一的对象中心表示。将MUFASA集成到现有OCL方法中，显著提升了多个数据集上的分割性能，达到新基准，同时改善训练收敛性，仅带来微小的推理开销。
原文摘要 · Abstract (English)
Unsupervised object-centric learning (OCL) decomposes visual scenes into distinct entities. Slot attention is a popular approach that represents individual objects as latent vectors, called slots. Current methods obtain these slot representations solely from the last layer of a pre-trained vision transformer (ViT), ignoring valuable, semantically rich information encoded across the other layers. To better utilize this latent semantic information, we introduce MUFASA, a lightweight plug-and-play framework for slot-attention-based approaches to unsupervised object segmentation. Our model computes slot attention across multiple feature layers of the ViT encoder, fully leveraging their semantic richness. We propose a fusion strategy to aggregate slots obtained on multiple layers into a unified object-centric representation. Integrating MUFASA into existing OCL methods improves their segmentation results across multiple datasets, setting a new state of the art while simultaneously improving training convergence with only minor inference overhead.
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。