Multimodal Retrieval-Augmented Generation (MRAG) has recently been explored to empower Large Vision Language Models (LVLMs) with more comprehensive and up-to-date contextual knowledge, aiming to compensate for their limited and coarse-grained parametric knowledge in knowledge-intensive tasks. However, the retrieved contextual knowledge is usually not aligned with LVLMs’ internal parametric knowledge, leading to knowledge conflicts and further unreliable responses. To tackle this issue, we design KCM, a training-free and plug-and-play framework that can effectively mitigate knowledge conflicts while incorporating MRAG for more accurate LVLM responses. KCM enhances contextual knowledge utilization by modifying the LVLM architecture from three key perspectives. First, KCM adaptively adjusts attention distributions among multiple attention heads, encouraging LVLMs to focus on contextual knowledge with reduced distraction. Second, KCM identifies and prunes knowledge-centric LVLM neurons that encode coarse-grained parametric knowledge, thereby suppressing interferences and enabling more effective integration of contextual knowledge. Third, KCM amplifies the information flow from the input context by injecting supplementary context logits, reinforcing its contribution to the final output. Extensive experiments over multiple LVLMs and benchmarks show that KCM outperforms the state-of-the-art consistently by large margins, incurring neither extra training nor external tools.

Image fusion aims to synthesize a single high-quality image from a pair of inputs captured under challenging conditions, such as differing exposure levels or focal depths. A core challenge lies in effectively handling disparities in dynamic range and focus depth between the inputs. With the advent of vision–language models, recent methods incorporate textual descriptions as auxiliary guidance to enhance fusion quality. However, simply incorporating coarse-grained descriptions hampers the understanding of fine-grained details and poses challenges for precise cross-modal alignment. To address these limitations, we propose Multi-grained Text-guided Image Fusion (MTIF), a novel fusion paradigm with three key designs. First, it introduces multi-grained textual descriptions that separately capture fine details, structural cues, and semantic content, guiding image fusion through a hierarchical cross-modal modulation module. Second, it involves supervision signals at each granularity to facilitate alignment between visual and textual features and enhance the utility of auxiliary text. Third, it adopts a saliency-driven enrichment module to augment training data with dense semantic content, further strengthening the cross-modal modulation and alignment. Extensive experiments show that MTIF consistently outperforms previous methods on both multi-exposure and multi-focus image fusion tasks.

Generalized Category Discovery is a significant and complex task that aims to identify both known and undefined novel categories from a set of unlabeled data, leveraging another labeled dataset containing only known categories. The primary challenges stem from model bias induced by pre-training on only known categories and the lack of precise supervision for novel ones, leading to category bias towards known categories and category confusion among different novel categories, which hinders models' ability to identify novel categories effectively. To address these challenges, we propose a novel framework named Self-Debiasing Calibration (SDC). Unlike prior methods that regard model bias towards known categories as an obstacle to novel category identification, SDC provides a novel insight into unleashing the potential of the bias to facilitate novel category learning. Specifically, we utilize the biased pre-trained model to guide the subsequent learning process on unlabeled data. The output of the biased model serves two key purposes. First, it provides an accurate modeling of category bias, which can be utilized to measure the degree of bias and debias the output of the current training model. Second, it offers valuable insights for distinguishing different novel categories by transferring knowledge between similar categories. Based on these insights, SDC dynamically adjusts the output logits of the current training model using the output of the biased model. This approach produces less biased logits to effectively address the issue of category bias towards known categories, and generates more accurate pseudo labels for unlabeled data, thereby mitigating category confusion for novel categories. Experiments on three benchmark datasets show that SDC outperforms SOTA methods, especially in the identification of novel categories.

Despite great success across various multimodal tasks, Large Vision-Language Models (LVLMs) often encounter object hallucinations with generated textual responses being inconsistent with the actual objects in images. We examine different LVLMs and pinpoint that one root cause of object hallucinations lies with deficient attention on discriminative image features. Specifically, LVLMs often predominantly attend to prompt-irrelevant global features instead of prompt-relevant local features, undermining their visual grounding capacity and leading to object hallucinations. We propose Assembly of Global and Local Attention (AGLA), a training-free and plug-and-play approach that mitigates hallucinations by assembling global features for response generation and local features for visual discrimination simultaneously. Specifically, we introduce an image-prompt matching scheme that captures prompt-relevant local features from images, leading to an augmented view of the input image where prompt-relevant content is highlighted while irrelevant distractions are suppressed. Hallucinations can thus be mitigated with a calibrated logit distribution that is from generative global features of the original image and discriminative local features of the augmented image. Extensive experiments show the superiority of AGLA in LVLM hallucination mitigation, demonstrating its wide applicability across both discriminative and generative tasks. Our code is available at https://github.com/Lackel/AGLA.

Despite their recent progress, Multimodal Large Language Models (MLLMs) often struggle in knowledge-intensive tasks due to the limited and outdated parametric knowledge acquired during training. Multimodal Retrieval Augmented Generation addresses this issue by retrieving contextual knowledge from external databases, thereby enhancing MLLMs with expanded knowledge sources. However, existing MLLMs often fail to fully leverage the retrieved contextual knowledge for response generation. We examine representative MLLMs and identify two major causes, namely, attention bias toward different tokens and knowledge conflicts between parametric and contextual knowledge. To this end, we design Adaptive Logits Fusion and Attention Reallocation (ALFAR), a training-free and plug-and-play approach that improves MLLM responses by maximizing the utility of the retrieved knowledge. Specifically, ALFAR tackles the challenges from two perspectives. First, it alleviates attention bias by adaptively shifting attention from visual tokens to relevant context tokens according to query-context relevance. Second, it decouples and weights parametric and contextual knowledge at output logits, mitigating conflicts between the two types of knowledge. As a plug-and-play method, ALFAR achieves superior performance across diverse datasets without requiring additional training or external tools. Extensive experiments over multiple MLLMs and benchmarks show that ALFAR consistently outperforms the state-of-the-art by large margins. Our code and data are available at https://github.com/Lackel/ALFAR.

Cross-Domain Few-Shot Segmentation (CD-FSS) poses the challenge of segmenting novel categories from a distinct domain using only limited exemplars. In this paper, we undertake a comprehensive study of CD-FSS and uncover two crucial insights: (i) the necessity of a fine-tuning stage to effectively transfer the learned meta-knowledge across domains, and (ii) the overfitting risk during the naive fine-tuning due to the scarcity of novel category examples. With these insights, we propose a novel cross-domain fine-tuning strategy that addresses the challenging CD-FSS tasks. We first design Bi-directional Few-shot Prediction (BFP), which establishes support-query correspondence in bi-directional manner, crafting augmented supervision to reduce the overfitting risk. Then we further extend BFP into Iterative Few-shot Adaptor (IFA), which is a recursive framework to capture the support-query correspondence iteratively, targeting maximal exploitation of supervisory signals from the sparse novel category samples. Extensive empirical evaluations show that our method significantly outperforms the state-of-the-arts (+7.8%), which verifies that IFA tackles the cross-domain challenges and mitigates the overfitting simultaneously.

The primary color profile of the same identity is assumed to remain consistent in typical Person Re-identification (Person ReID) tasks. However, this assumption may be invalid in real-world situations and images hold variant color profiles, because of cross-modality cameras or identity with different clothing. To address this issue, we propose Color Space Learning (CSL) for those Cross-Color Person ReID problems. Specifically, CSL guides the model to be less color-sensitive with two modules: Image-level Color-Augmentation and Pixel-level Color-Transformation. The first module increases the color diversity of the inputs and guides the model to focus more on the non-color information. The second module projects every pixel of input images onto a new color space. In addition, we introduce a new Person ReID benchmark across RGB and Infrared modalities, NTU-Corridor, which is the first with privacy agreements from all participants. To evaluate the effectiveness and robustness of our proposed CSL, we evaluate it on several Cross-Color Person ReID benchmarks. Our method surpasses the state-of-the-art methods consistently. The code and benchmark are available at: https://github.com/niejiahao1998/CSL

Continual learning (CL) with Vision-Language Models (VLMs) has overcome the constraints of traditional CL, which only focuses on previously encountered classes. During the CL of VLMs, we need not only to prevent the catastrophic forgetting on incrementally learned knowledge but also to preserve the zero-shot ability of VLMs. However, existing methods require additional reference datasets to maintain such zero-shot ability and rely on domain-identity hints to classify images across different domains. In this study, we propose Regression-based Analytic Incremental Learning (RAIL), which utilizes a recursive ridge regression-based adapter to learn from a sequence of domains in a non-forgetting manner and decouple the cross-domain correlations by projecting features to a higher-dimensional space. Cooperating with a training-free fusion module, RAIL absolutely preserves the VLM's zero-shot ability on unseen domains without any reference data. Additionally, we introduce Cross-domain Task-Agnostic Incremental Learning (X-TAIL) setting. In this setting, a CL learner is required to incrementally learn from multiple domains and classify test images from both seen and unseen domains without any domain-identity hint. We theoretically prove RAIL's absolute memorization on incrementally learned domains. Experiment results affirm RAIL's state-of-the-art performance in both X-TAIL and existing Multi-domain Task-Incremental Learning settings. The code is released at https://github.com/linghan1997/Regression-based-Analytic-Incremental-Learning.

Vision-Language Pre-training has demonstrated its remarkable zero-shot recognition ability and potential to learn generalizable visual representations from languagesupervision. Taking a step ahead, language-supervised semantic segmentation enables spatial localization of textual inputs by learning pixel grouping solely from image-text pairs. Nevertheless, the state-of-the-art suffers from a clear semantic gap between visual and textual modalities: plenty of visual concepts appeared in images are missing in their paired captions. Such semantic misalignment circulates in pre-training, leading to inferior zero-shot performance in dense predictions due to insufficient visual concepts captured in textual representations. To close such semantic gap, we propose Concept Curation (CoCu), a pipeline that leverages CLIP to compensate for the missing semantics. For each image-text pair, we establish a concept archive that maintains potential visually-matched concepts with our proposed vision-driven expansion and text-to-vision-guided ranking. Relevant concepts can thus be identified via cluster-guided sampling and fed into pre-training, thereby bridging the gap between visual and textual semantics. Extensive experiments over a broad suite of 8 segmentation benchmarks show that CoCu achieves superb zero-shot transfer performance and greatly boosts language-supervised segmentation baseline by a large margin, suggesting the value of closing semantic gap in pre-training data.