Diffusion-Based Cross-Modal Feature Extraction for Multi-Label Classification

Multi-label classification has broad applications and depends on powerful representations capable of capturing multi-label interactions. We introduce \textit{Diff-Feat}, a simple but powerful framework that extracts intermediate features from pre-trained diffusion-Transformer models for images and text, and fuses them for downstream tasks. We observe that for vision tasks, the most discriminative intermediate feature along the diffusion process occurs at the middle step and is located in the middle block in Transformer. In contrast, for language tasks, the best feature occurs at the noise-free step and is located in the deepest block. In particular, we observe a striking phenomenon across varying datasets: a mysterious "Layer $12$" consistently yields the best performance on various downstream classification tasks for images (under DiT-XL/2-256$\times$256). We devise a heuristic local-search algorithm that pinpoints the locally optimal "image-text"$\times$"block-timestep" pair among a few candidates, avoiding an exhaustive grid search. A simple fusion-linear projection followed by addition-of the selected representations yields state-of-the-art performance: 98.6\% mAP on MS-COCO-enhanced and 45.7\% mAP on Visual Genome 500, surpassing strong CNN, graph, and Transformer baselines by a wide margin. t-SNE and clustering metrics further reveal that \textit{Diff-Feat} forms tighter semantic clusters than unimodal counterparts. The code is available at https://github.com/lt-0123/Diff-Feat.

ProtoMedX: Towards Explainable Multi-Modal Prototype Learning for Bone Health Classification

Bone health studies are crucial in medical practice for the early detection and treatment of Osteopenia and Osteoporosis. Clinicians usually make a diagnosis based on densitometry (DEXA scans) and patient history. The applications of AI in this field are ongoing research. Most successful methods rely on deep learning models that use vision alone (DEXA/X-ray imagery) and focus on prediction accuracy, while explainability is often disregarded and left to post hoc assessments of input contributions. We propose ProtoMedX, a multi-modal model that uses both DEXA scans of the lumbar spine and patient records. ProtoMedX's prototype-based architecture is explainable by design, which is crucial for medical applications, especially in the context of the upcoming EU AI Act, as it allows explicit analysis of model decisions, including incorrect ones. ProtoMedX demonstrates state-of-the-art performance in bone health classification while also providing explanations that can be visually understood by clinicians. Using a dataset of 4,160 real NHS patients, the proposed ProtoMedX achieves 87.58% accuracy in vision-only tasks and 89.8% in its multi-modal variant, both surpassing existing published methods.

LLM4MG: Adapting Large Language Model for Multipath Generation via Synesthesia of Machines

Based on Synesthesia of Machines (SoM), a large language model (LLM) is adapted for multipath generation (LLM4MG) for the first time. Considering a typical sixth-generation (6G) vehicle-to-infrastructure (V2I) scenario, a new multi-modal sensing-communication dataset is constructed, named SynthSoM-V2I, including channel multipath information, millimeter wave (mmWave) radar sensory data, RGB-D images, and light detection and ranging (LiDAR) point clouds. Based on the SynthSoM-V2I dataset, the proposed LLM4MG leverages Large Language Model Meta AI (LLaMA) 3.2 for multipath generation via multi-modal sensory data. The proposed LLM4MG aligns the multi-modal feature space with the LLaMA semantic space through feature extraction and fusion networks. To further achieve general knowledge transfer from the pre-trained LLaMA for multipath generation via multi-modal sensory data, the low-rank adaptation (LoRA) parameter-efficient fine-tuning and propagation-aware prompt engineering are exploited. Simulation results demonstrate that the proposed LLM4MG outperforms conventional deep learning-based methods in terms of line-of-sight (LoS)/non-LoS (NLoS) classification with accuracy of 92.76%, multipath power/delay generation precision with normalized mean square error (NMSE) of 0.099/0.032, and cross-vehicular traffic density (VTD), cross-band, and cross-scenario generalization. The utility of the proposed LLM4MG is validated by real-world generalization. The necessity of high-precision multipath generation for system design is also demonstrated by channel capacity comparison.

No Modality Left Behind: Adapting to Missing Modalities via Knowledge Distillation for Brain Tumor Segmentation

Accurate brain tumor segmentation is essential for preoperative evaluation and personalized treatment. Multi-modal MRI is widely used due to its ability to capture complementary tumor features across different sequences. However, in clinical practice, missing modalities are common, limiting the robustness and generalizability of existing deep learning methods that rely on complete inputs, especially under non-dominant modality combinations. To address this, we propose AdaMM, a multi-modal brain tumor segmentation framework tailored for missing-modality scenarios, centered on knowledge distillation and composed of three synergistic modules. The Graph-guided Adaptive Refinement Module explicitly models semantic associations between generalizable and modality-specific features, enhancing adaptability to modality absence. The Bi-Bottleneck Distillation Module transfers structural and textural knowledge from teacher to student models via global style matching and adversarial feature alignment. The Lesion-Presence-Guided Reliability Module predicts prior probabilities of lesion types through an auxiliary classification task, effectively suppressing false positives under incomplete inputs. Extensive experiments on the BraTS 2018 and 2024 datasets demonstrate that AdaMM consistently outperforms existing methods, exhibiting superior segmentation accuracy and robustness, particularly in single-modality and weak-modality configurations. In addition, we conduct a systematic evaluation of six categories of missing-modality strategies, confirming the superiority of knowledge distillation and offering practical guidance for method selection and future research. Our source code is available at https://github.com/Quanato607/AdaMM.

M-PACE: Mother Child Framework for Multimodal Compliance

Ensuring that multi-modal content adheres to brand, legal, or platform-specific compliance standards is an increasingly complex challenge across domains. Traditional compliance frameworks typically rely on disjointed, multi-stage pipelines that integrate separate modules for image classification, text extraction, audio transcription, hand-crafted checks, and rule-based merges. This architectural fragmentation increases operational overhead, hampers scalability, and hinders the ability to adapt to dynamic guidelines efficiently. With the emergence of Multimodal Large Language Models (MLLMs), there is growing potential to unify these workflows under a single, general-purpose framework capable of jointly processing visual and textual content. In light of this, we propose Multimodal Parameter Agnostic Compliance Engine (M-PACE), a framework designed for assessing attributes across vision-language inputs in a single pass. As a representative use case, we apply M-PACE to advertisement compliance, demonstrating its ability to evaluate over 15 compliance-related attributes. To support structured evaluation, we introduce a human-annotated benchmark enriched with augmented samples that simulate challenging real-world conditions, including visual obstructions and profanity injection. M-PACE employs a mother-child MLLM setup, demonstrating that a stronger parent MLLM evaluating the outputs of smaller child models can significantly reduce dependence on human reviewers, thereby automating quality control. Our analysis reveals that inference costs reduce by over 31 times, with the most efficient models (Gemini 2.0 Flash as child MLLM selected by mother MLLM) operating at 0.0005 per image, compared to 0.0159 for Gemini 2.5 Pro with comparable accuracy, highlighting the trade-off between cost and output quality achieved in real time by M-PACE in real life deployment over advertising data.

WHU-STree: A Multi-modal Benchmark Dataset for Street Tree Inventory

Street trees are vital to urban livability, providing ecological and social benefits. Establishing a detailed, accurate, and dynamically updated street tree inventory has become essential for optimizing these multifunctional assets within space-constrained urban environments. Given that traditional field surveys are time-consuming and labor-intensive, automated surveys utilizing Mobile Mapping Systems (MMS) offer a more efficient solution. However, existing MMS-acquired tree datasets are limited by small-scale scene, limited annotation, or single modality, restricting their utility for comprehensive analysis. To address these limitations, we introduce WHU-STree, a cross-city, richly annotated, and multi-modal urban street tree dataset. Collected across two distinct cities, WHU-STree integrates synchronized point clouds and high-resolution images, encompassing 21,007 annotated tree instances across 50 species and 2 morphological parameters. Leveraging the unique characteristics, WHU-STree concurrently supports over 10 tasks related to street tree inventory. We benchmark representative baselines for two key tasks--tree species classification and individual tree segmentation. Extensive experiments and in-depth analysis demonstrate the significant potential of multi-modal data fusion and underscore cross-domain applicability as a critical prerequisite for practical algorithm deployment. In particular, we identify key challenges and outline potential future works for fully exploiting WHU-STree, encompassing multi-modal fusion, multi-task collaboration, cross-domain generalization, spatial pattern learning, and Multi-modal Large Language Model for street tree asset management. The WHU-STree dataset is accessible at: https://github.com/WHU-USI3DV/WHU-STree.

Surformer v2: A Multimodal Classifier for Surface Understanding from Touch and Vision

Multimodal surface material classification plays a critical role in advancing tactile perception for robotic manipulation and interaction. In this paper, we present Surformer v2, an enhanced multi-modal classification architecture designed to integrate visual and tactile sensory streams through a late(decision level) fusion mechanism. Building on our earlier Surformer v1 framework [1], which employed handcrafted feature extraction followed by mid-level fusion architecture with multi-head cross-attention layers, Surformer v2 integrates the feature extraction process within the model itself and shifts to late fusion. The vision branch leverages a CNN-based classifier(Efficient V-Net), while the tactile branch employs an encoder-only transformer model, allowing each modality to extract modality-specific features optimized for classification. Rather than merging feature maps, the model performs decision-level fusion by combining the output logits using a learnable weighted sum, enabling adaptive emphasis on each modality depending on data context and training dynamics. We evaluate Surformer v2 on the Touch and Go dataset [2], a multi-modal benchmark comprising surface images and corresponding tactile sensor readings. Our results demonstrate that Surformer v2 performs well, maintaining competitive inference speed, suitable for real-time robotic applications. These findings underscore the effectiveness of decision-level fusion and transformer-based tactile modeling for enhancing surface understanding in multi-modal robotic perception.

SimCroP: Radiograph Representation Learning with Similarity-driven Cross-granularity Pre-training

Medical vision-language pre-training shows great potential in learning representative features from massive paired radiographs and reports. However, in computed tomography (CT) scans, the distribution of lesions which contain intricate structures is characterized by spatial sparsity. Besides, the complex and implicit relationships between different pathological descriptions in each sentence of the report and their corresponding sub-regions in radiographs pose additional challenges. In this paper, we propose a Similarity-Driven Cross-Granularity Pre-training (SimCroP) framework on chest CTs, which combines similarity-driven alignment and cross-granularity fusion to improve radiograph interpretation. We first leverage multi-modal masked modeling to optimize the encoder for understanding precise low-level semantics from radiographs. Then, similarity-driven alignment is designed to pre-train the encoder to adaptively select and align the correct patches corresponding to each sentence in reports. The cross-granularity fusion module integrates multimodal information across instance level and word-patch level, which helps the model better capture key pathology structures in sparse radiographs, resulting in improved performance for multi-scale downstream tasks. SimCroP is pre-trained on a large-scale paired CT-reports dataset and validated on image classification and segmentation tasks across five public datasets. Experimental results demonstrate that SimCroP outperforms both cutting-edge medical self-supervised learning methods and medical vision-language pre-training methods. Codes and models are available at https://github.com/ToniChopp/SimCroP.

EHWGesture -- A dataset for multimodal understanding of clinical gestures

Hand gesture understanding is essential for several applications in human-computer interaction, including automatic clinical assessment of hand dexterity. While deep learning has advanced static gesture recognition, dynamic gesture understanding remains challenging due to complex spatiotemporal variations. Moreover, existing datasets often lack multimodal and multi-view diversity, precise ground-truth tracking, and an action quality component embedded within gestures. This paper introduces EHWGesture, a multimodal video dataset for gesture understanding featuring five clinically relevant gestures. It includes over 1,100 recordings (6 hours), captured from 25 healthy subjects using two high-resolution RGB-Depth cameras and an event camera. A motion capture system provides precise ground-truth hand landmark tracking, and all devices are spatially calibrated and synchronized to ensure cross-modal alignment. Moreover, to embed an action quality task within gesture understanding, collected recordings are organized in classes of execution speed that mirror clinical evaluations of hand dexterity. Baseline experiments highlight the dataset's potential for gesture classification, gesture trigger detection, and action quality assessment. Thus, EHWGesture can serve as a comprehensive benchmark for advancing multimodal clinical gesture understanding.

Singular Value Few-shot Adaptation of Vision-Language Models

Vision-language models (VLMs) like CLIP have shown impressive zero-shot and few-shot learning capabilities across diverse applications. However, adapting these models to new fine-grained domains remains difficult due to reliance on prompt engineering and the high cost of full model fine-tuning. Existing adaptation approaches rely on augmented components, such as prompt tokens and adapter modules, which could limit adaptation quality, destabilize the model, and compromise the rich knowledge learned during pretraining. In this work, we present \textbf{CLIP-SVD}, a novel \textit{multi-modal} and \textit{parameter-efficient} adaptation technique that leverages Singular Value Decomposition (SVD) to modify the internal parameter space of CLIP without injecting additional modules. Specifically, we fine-tune only the singular values of the CLIP parameter matrices to rescale the basis vectors for domain adaptation while retaining the pretrained model. This design enables enhanced adaptation performance using only \textbf{0.04\%} of the model's total parameters and better preservation of its generalization ability. CLIP-SVD achieves state-of-the-art classification results on 11 natural and 10 biomedical datasets, outperforming previous methods in both accuracy and generalization under few-shot settings. Additionally, we leverage a natural language-based approach to analyze the effectiveness and dynamics of the CLIP adaptation to allow interpretability of CLIP-SVD. The code is publicly available at https://github.com/HealthX-Lab/CLIP-SVD.