Abstract
Meta-learning aims to uniformly sample homogeneous support-query pairs, characterized by the same categories and similar attributes, and extract useful inductive biases through identical network architectures. However, this identical network design results in over-semantic homogenization. To address this, we propose a novel homologous but heterogeneous network. By treating support-query pairs as dual perspectives, we introduce heterogeneous visual aggregation (HA) modules to enhance complementarity while preserving semantic commonality. To further reduce semantic noise and amplify the uniqueness of heterogeneous semantics, we design a heterogeneous transfer (HT) module. Finally, we propose heterogeneous CLIP (HC) textual information to enhance the generalization capability of multimodal models. In the weakly-supervised few-shot semantic segmentation (WFSS) task, with only 1/24 of the parameters of existing state-of-the-art models, TLG achieves a 13.2\% improvement on Pascal-5\textsuperscript{i} and a 9.7\% improvement on COCO-20\textsuperscript{i}. To the best of our knowledge, TLG is also the first weakly supervised (image-level) model that outperforms fully supervised (pixel-level) models under the same backbone architectures. The code is available at this https URL.
Abstract (translated)
元学习的目标是均匀采样同质的支持-查询对，这些对具有相同的类别和相似的属性，并通过相同的网络架构提取有用的归纳偏差。然而，这种相同的设计导致了过度语义同质化。为了解决这个问题，我们提出了一种新型但异构的网络设计。通过将支持-查询对视为双重视角，我们引入了异构视觉聚合（HA）模块来增强互补性同时保留语义共性。为了进一步减少语义噪声并放大异构语义的独特性，我们设计了一个异构传输（HT）模块。最后，我们提出了一种异构CLIP (HC) 文本信息以提升多模态模型的泛化能力。 在弱监督下的少样本语义分割(WFSS)任务中，TLG模型仅使用现有最先进模型参数量的1/24，在Pascal-5i上实现了13.2%的改进，并在COCO-20i上实现了9.7%的改进。据我们所知，TLG也是第一个超越完全监督（像素级别）模型的弱监督（图像级别）模型，在相同的骨干架构下表现出色。 代码可在提供的链接处获取：[此URL](https://this-url.com) (请将"this https URL"替换为实际可用的具体URL地址)。
URL
https://arxiv.org/abs/2508.16159
