Vision-Language Retrieval (VLR) aims to retrieve relevant visual or textual information from multimodal data using language or image queries. However, traditional VLR methods often rely on data-driven shallow semantic alignment and fail to understand the deeper structural and fine-grained entity features of queries, resulting in poor performance on multi-entity layouts and challenging entities. In this paper, we propose the Layout-Aware and Sketch-Enhanced (LASE) VLR framework, which refines query representations by incorporating multimodal layout and sketch knowledge. Specifically, layout knowledge encodes the spatial arrangement of entities, while sketch knowledge refines entity perception by capturing essential structural details. To extract these knowledge representations, we leverage Large Language Models' (LLMs) powerful semantic understanding for layout generation, and Diffusion Models' (DMs) fine-grained cross-modal generative capabilities for sketch generation. However, integrating knowledge into queries may introduce biases and query-specific preferences due to varying visual content and knowledge demands. To address this, we propose the Gated Dual-Stream Knowledge Module (GDKM), which consists of a multi-instance fusion network with a sample-aware gating network. The fusion network aggregates diverse knowledge using multi-head attention to reduce bias, while the gating network adjusts knowledge weights based on query characteristics. Extensive experiments demonstrate that the LASE significantly enhances VLR performance across multiple benchmarks, with superior generalization and transferability.

Vision-language retrieval (VLR), which uses text or image queries to retrieve corresponding cross-modal content, plays a crucial role in multimedia and computer vision tasks. However, challenging concepts in queries often confuse retrievers, limiting their ability to align concepts with visual content. Existing query optimization methods neglect retrievers’ preferences (i.e., text descriptions that better match their corresponding visual content), resulting in unadapted to the retriever and leading to suboptimal performance. To address this, we propose the Retriever-Adaptive Query Optimization (RAQO), an interpretable framework that rewrites queries based on retriever-specific preferences. Specifically, we first leverages multimodal large language Models (MLLMs) and retrieval's feedback to construct the MLLMs-Driven Preference-Aware Dataset Engine (MPADE), which automatically refine queries offline, capturing the retriever’s implicit preferences. Then, we introduce a ``detect-then-rewrite" chain-of-thought rewriting (ReCoT) strategy equipped with a progressive preference alignment pipeline, including three stages: ambiguity detection fine-tuning, query rewriting fine-tuning, and preference rank optimization. This design enables the rewriter to focus on confusing concepts and produce retriever-adapted, high-quality queries. Extensive VLR benchmark experiments have demonstrated the superiority of RAQO in cross-modal retrieval, as well as its interpretability, generalizability and transferability.

Composed Image Retrieval (CIR) enables image search by combining a reference image with modification text. Intrinsic noise in CIR triplets incurs intrinsic uncertainty and threatens model's robustness. Probabilistic learning approaches have shown promise in addressing such issues; however, they fall short for CIR due to their instance-level holistic modeling and homogeneous treatments for queries and targets. This paper introduces a Heterogeneous Uncertainty-Guided (HUG) paradigm to overcome these limitations. HUG utilizes a fine-grained probabilistic learning framework, where queries and targets are represented by Gaussian embeddings capturing detailed concepts and uncertainties. We customize heterogeneous uncertainty estimations for multi-modal queries and uni-modal targets. Given a query, we capture uncertainties not only regarding uni-modal content quality but also multi-modal coordination, followed by a provable dynamic weighting mechanism to derive the comprehensive query uncertainty. We further design uncertainty-guided objectives, including query-target holistic contrast and fine-grained contrasts with comprehensive negative sampling strategies, which effectively enhance discriminative learning. Experiments on benchmarks demonstrate HUG's effectiveness beyond state-of-the-art baselines, with faithful analysis justifying the technical contributions.

With the emergence of large multimodal models, dual-encoder alignment via contrastive learning has seen a resurgence. However, the escalating model size demands effective Parameter-Efficient Fine-Tuning (PEFT). While LoRA is a promising inference-free alternative to adapters, we find that its naive application to multimodal tasks causes a severe rank imbalance, favoring the text modality and FFN layers. To address this, we propose HALoRA (Hierarchical Allocation LoRA), which introduces a component-wise budget allocator to ensure balanced fine-tuning across both modalities and their internal components. This is complemented by a gradient-approximated initialization to accelerate convergence. With only half the parameters of adapters, HALoRA achieves superior or competitive performance in retrieval and zero-shot classification. Our work presents a more principled approach to multimodal LoRA, uncovering an intriguing asymmetry in vision-language alignment.

Vision-language retrieval (VLR) has attracted significant attention in both academia and industry, which involves using text (or images) as queries to retrieve corresponding images (or text). However, existing methods often neglect the rich visual semantics knowledge of entities, thus leading to incorrect retrieval results. To address this problem, we propose the Entity Visual Description enhanced CLIP (EvdCLIP), designed to leverage the visual knowledge of entities to enrich queries. Specifically, since humans recognize entities through visual cues, we employ a large language model (LLM) to generate Entity Visual Descriptions (EVDs) as alignment cues to complement textual data. These EVDs are then integrated into raw queries to create visually-rich, EVD-enhanced queries. Furthermore, recognizing that EVD-enhanced queries may introduce noise or low-quality expansions, we develop a novel, trainable EVD-aware Rewriter (EaRW) for vision-language retrieval tasks. EaRW utilizes EVD knowledge and the generative capabilities of the language model to effectively rewrite queries. With our specialized training strategy, EaRW can generate high-quality and low-noise EVD-enhanced queries. Extensive quantitative and qualitative experiments on image-text retrieval benchmarks validate the superiority of EvdCLIP on vision-language retrieval tasks.

Haomiao Tang, Jinpeng Wang, Yuang Peng, GuangHao Meng, Ruisheng Luo, Bin Chen, Long Chen, Yaowei Wang, Shu-Tao Xia. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Deep neural network (DNN) based scene text recognition (STR) methods usually require a large amount of annotated data for training, which is time-consuming and cost-expensive in practice. To address this issue, many data augmentation methods have been developed to train recognizers by improving the diversity of training samples. However, most existing methods neglect the difficulty inherent in samples, and easily suffer from the problem of over-diversity, i.e., the distribution of the augmented data significantly deviates from that of clean data. In this paper, we propose a novel difficulty-aware data augmentation framework for scene text recognition, which jointly considers the difficulty of samples and the strength of augmentations. Specifically, our framework first predicts the sample difficulty, followed by an adaptive data augmentation strategy. Furthermore, we build a more diverse set of augmentation methods for STR and integrate it into our augmentation framework. Extensive experiments on scene text recognition benchmarks show that our augmentation framework significantly improves the performance of recognizers.

Scene text image super-resolution (STISR), aiming to improve image quality while boosting downstream scene text recognition accuracy, has recently achieved great success. However, most existing methods treat the foreground (character regions) and background (non-character regions) equally in the forward process, and neglect the disturbance from the complex background, thus limiting the performance. To address these issues, in this paper, we propose a novel method LEMMA that explicitly models character regions to produce high-level text-specific guidance for super-resolution. To model the location of characters effectively, we propose the location enhancement module to extract character region features based on the attention map sequence. Besides, we propose the multi-modal alignment module to perform bidirectional visual-semantic alignment to generate high-quality prior guidance, which is then incorporated into the super-resolution branch in an adaptive manner using the proposed adaptive fusion module. Experiments on TextZoom and four scene text recognition benchmarks demonstrate the superiority of our method over other state-of-the-art methods. Code is available at https://github.com/csguoh/LEMMA.

Recognizing characters from low-resolution (LR) text images poses a significant challenge due to the information deficiency as well as the noise and blur in low-quality images. Current solutions for low-resolution text recognition (LTR) typically rely on a two-stage pipeline that involves super-resolution as the first stage followed by the second-stage recognition. Although this pipeline is straightforward and intuitive, it has to use an additional super-resolution network, which causes inefficiencies during training and testing. Moreover, the recognition accuracy of the second stage heavily depends on the reconstruction quality of the first stage, causing ineffectiveness.In this work, we attempt to address these challenges from a novel perspective: adapting the recognizer to low-resolution inputs by transferring the knowledge from the high-resolution. Guided by this idea, we propose an efficient and effective knowledge distillation framework to achieve multi-level knowledge transfer.Specifically, the visual focus loss is proposed to extract the character position knowledge with resolution gap reduction and character region focus, the semantic contrastive loss is employed to exploit the contextual semantic knowledge with contrastive learning, and the soft logits loss facilitates both local word-level and global sequence-level learning from the soft teacher label.Extensive experiments show that the proposed one-stage pipeline significantly outperforms super-resolution based two-stage frameworks in terms of effectiveness and efficiency, accompanied by favorable robustness.Code is available at https://github.com/csguoh/KD-LTR.

Deep neural networks (DNNs) have been widely used in scene text recognition, and achieved remarkable performance. Such DNN-based scene text recognizers usually require plenty of training data for training, but data collection and annotation is usually cost-expensive in practice. To alleviate this issue, data augmentation is often applied to train the scene text recognizers. However, existing data augmentation methods including affine transformation and elastic transformation methods suffer from the problems of under- and over-diversity, due to the complexity of text contents and shapes. In this paper, we propose a sample-aware data augmentor to transform samples adaptively based on the contents of samples. Specifically, our data augmentor consists of three parts: gated module, affine transformation module, and elastic transformation module. In our data augmentor, affine transformation module focuses on keeping the affinity of samples, while elastic transformation module aims to improve the diversity of samples. With the gated module, our data augmentor determines transformation type adaptively based on the properties of training samples and the recognizer capability during the training process. Besides, our framework introduces an adversarial learning strategy to optimize the augmentor and the recognizer jointly. Extensive experiments on scene text recognition benchmarks show that our sample-aware data augmentor significantly improves the performance of state-of-the-art scene text recognizer.

Scene text detection (STD) aims to locate text in images and plays an important role in many computer vision tasks including automatic driving and text recognition systems. Recently, deep neural networks (DNNs) have been widely and successfully used in scene text detection, leading to plenty of DNN-based STD methods including regression-based and segmentation-based STD methods. However, recent studies have also shown that DNN is vulnerable to adversarial attacks, which can significantly degrade the performance of DNN models. In this paper, we investigate the robustness of DNN-based STD methods against adversarial attacks. To this end, we propose a generic and efficient attack method to generate adversarial examples, which are produced by adding small but imperceptible adversarial perturbation to the input images. Experiments on attacking four various models and a real-world STD engine of Google optical character recognition (OCR) show that the state-of-the-art DNN-based STD methods including regression-based and segmentation-based methods are vulnerable to adversarial attacks.