Fine-grained Retrieval Prompt Tuning

TL;DR

This paper introduces Fine-grained Retrieval Prompt Tuning (FRPT), a method that leverages prompt learning and feature adaptation to improve fine-grained object retrieval without full model fine-tuning, achieving state-of-the-art results.

Contribution

FRPT proposes a novel approach using discriminative perturbation prompts and category-specific awareness head to enhance fine-grained retrieval with fewer learnable parameters.

Findings

Achieves state-of-the-art performance on three fine-grained datasets.

Reduces the number of parameters needed for effective fine-grained retrieval.

Maintains generalization and discrimination of representations.

Abstract

Fine-grained object retrieval aims to learn discriminative representation to retrieve visually similar objects. However, existing top-performing works usually impose pairwise similarities on the semantic embedding spaces or design a localization sub-network to continually fine-tune the entire model in limited data scenarios, thus resulting in convergence to suboptimal solutions. In this paper, we develop Fine-grained Retrieval Prompt Tuning (FRPT), which steers a frozen pre-trained model to perform the fine-grained retrieval task from the perspectives of sample prompting and feature adaptation. Specifically, FRPT only needs to learn fewer parameters in the prompt and adaptation instead of fine-tuning the entire model, thus solving the issue of convergence to suboptimal solutions caused by fine-tuning the entire model. Technically, a discriminative perturbation prompt (DPP) is introduced and deemed as a sample prompting process, which amplifies and even exaggerates some discriminative elements contributing to category prediction via a content-aware inhomogeneous sampling operation. In this way, DPP can make the fine-grained retrieval task aided by the perturbation prompts close to the solved task during the original pre-training. Thereby, it preserves the generalization and discrimination of representation extracted from input samples. Besides, a category-specific awareness head is proposed and regarded as feature adaptation, which removes the species discrepancies in features extracted by the pre-trained model using category-guided instance normalization. And thus, it makes the optimized features only include the discrepancies among subcategories. Extensive experiments demonstrate that our FRPT with fewer learnable parameters achieves the state-of-the-art performance on three widely-used fine-grained datasets.