提出多头持续学习框架，实现细粒度服装检索的低成本增量更新。
A Multihead Continual Learning Framework for Fine-Grained Fashion Image Retrieval with Contrastive Learning and Exponential Moving Average Distillation
- 采用多头结构与对比学习，支持新类持续加入
- 相比静态方法仅需30%训练成本，性能相当
- 适合动态场景下的服装图像检索应用
现有细粒度服装图像检索（FIR）方法多假设场景静态，新增属性需全量重训，成本高昂。虽预训练模型支持零样本推理，但缺乏监督时准确率下降，且尚未有研究探索细粒度FIR的类增量学习（CIL）。本文提出多头持续学习框架MCL-FIR，结合对比学习与指数移动平均（EMA）知识蒸馏。该框架采用多头设计适应类别增量，将三元组输入重构为双元组并使用InfoNCE损失，提升训练效率；通过EMA蒸馏实现知识高效迁移。在四个数据集上的实验表明，MCL-FIR不仅具备良好可扩展性，还实现了效率与精度的优良平衡：在相近训练成本下显著优于现有CIL基线，相较静态方法性能相当，但训练成本仅为其约30%。源代码已公开于https://github.com/Dr-LingXiao/MCL-FIR。
原文摘要 · Abstract (English)
Most fine-grained fashion image retrieval (FIR) methods assume a static setting, requiring full retraining when new attributes appear, which is costly and impractical for dynamic scenarios. Although pretrained models support zero-shot inference, their accuracy drops without supervision, and no prior work explores class-incremental learning (CIL) for fine-grained FIR. We propose a multihead continual learning framework for fine-grained fashion image retrieval with contrastive learning and exponential moving average (EMA) distillation (MCL-FIR). MCL-FIR adopts a multi-head design to accommodate evolving classes across increments, reformulates triplet inputs into doublets with InfoNCE for simpler and more effective training, and employs EMA distillation for efficient knowledge transfer. Experiments across four datasets demonstrate that, beyond its scalability, MCL-FIR achieves a strong balance between efficiency and accuracy. It significantly outperforms CIL baselines under similar training cost, and compared with static methods, it delivers comparable performance while using only about 30% of the training cost. The source code is publicly available in https://github.com/Dr-LingXiao/MCL-FIR.
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。