Visible-Infrared Person Re-Identification (VI-ReID) is a challenging cross-modal matching task due to significant modality discrepancies. While current methods mainly focus on learning modality-invariant features through unified embedding spaces, they often focus solely on the common discriminative semantics across modalities while disregarding the critical role of modality-specific identity-aware knowledge in discriminative feature learning. To bridge this gap, we propose a novel Identity Clue Refinement and Enhancement (ICRE) network to mine and utilize the implicit discriminative knowledge inherent in modality-specific attributes. Initially, we design a Multi-Perception Feature Refinement (MPFR) module that aggregates shallow features from shared branches, aiming to capture modality-specific attributes that are easily overlooked. Then, we propose a Semantic Distillation Cascade Enhancement (SDCE) module, which distills identity-aware knowledge from the aggregated shallow features and guide the learning of modality-invariant features. Finally, an Identity Clues Guided (ICG) Loss is proposed to alleviate the modality discrepancies within the enhanced features and promote the learning of a diverse representation space. Extensive experiments across multiple public datasets clearly show that our proposed ICRE outperforms existing SOTA methods.

In this paper, we study the problem of shape part retrieval in the point cloud domain. Existing shape retrieval methods typically rely on the presence of a complete query object, but what if the part of interest is missing? We present the Part Retrieval Pipeline (PReP), which combines metric learning techniques with a trained classification model to evaluate the suitability of potential replacement parts from a database, within an application scenario aimed at circular economy. Through a progressively more difficult training procedure, PReP learns to recognize suitable parts based solely on shape context. Owing to its compact parameterization and low computational requirements, it can search a repository of tens of thousands of spare parts in just a few seconds. We also establish an alternative baseline approach for comparison, document the unique challenges associated with this task, and identify key design choices to address them.

Transmitting images through narrow-bandwidth underwater acoustic channels presents a significant challenge. However, for many applications, transmitting the entire image is not necessary. Instead, conveying only the critical information is typically sufficient to ensure task success. To address this, we have introduced a preprocessing strategy at the image acquisition stage, which involves down-sampling or feature extraction to distill the data into its essential components. This strategy is guided by utility considerations to ensure that the transmitted data is both concise and informative. However, existing quality evaluation methods are inadequate for providing accurate utility assessments and fail to simultaneously evaluate the utility of both images and features within a unified framework. To bridge this gap in underwater object detection (one of the typical underwater task), we have developed an underwater image utility dataset and introduced the Lightweight and Unified Utility for Underwater Image Evaluation (LU<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>IE). LU<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>IE leverages features derived from multiple detection models to enhance generalization. It then refines these features through dimensionality reduction and commonality extraction, and simplifies complexity through the use of knowledge distillation. Experimental results validate the effectiveness and scalability of LU<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>IE in accurately assessing the utility of underwater images.

Motion capture (MOCAP)-free music for performance generation using deep generative models is emerging as a promising solution for the next-generation animation industry. This technology allows users to create dynamic musical performance animations without relying on MOCAP. However, implementing MOCAP-free content-to-performance generation systems presents significant challenges. First, integrating various standalone models into a cohesive system is essential, as each model governs distinct aspects of the avatar's behavior. For example, a facial expression generation module influences the avatar's facial expressions, whereas a fingering generation module determines hand positions. Second, most applications focus on humanonly performance generation, such as virtual vocalists and virtual dancers, without considering interactions with other objects, such as instruments, referred to as human-instrument performance generation. To our knowledge, comprehensive human-instrument content-to-performance generation systems are still rare. In this paper, we present a complete content-to-performance generation system, demonstrating its capabilities through a web application that enables users to create animations from their music content. This system incorporates four modules to generate parameters for controlling avatars: 1) a facial expression module, 2) a fingering generation module, 3) a body movement generation module, and 4) a video shot generation module. Additionally, we integrate an expressive music synthesis module to generate expressive audio from symbolic music data. Quantitative evaluations confirm the effectiveness of the four modules, whereas a user study provides qualitative insights into the system's performance. The web application is available on our website (<uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://virtual-musician.iis.sinica.edu.tw:8800</uri>).

The development of Deep Neural Networks (DNNs) has enabled AI-driven models to excel in recognizing a limited set of classes within static environments. As AI systems progress, few-shot class-incremental learning (FSCIL) aims to expand their understanding of novel classes from minimal samples while retaining knowledge of previously encountered ones. However, most existing FSCIL models face significant challenges, including <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">inadequate adaptability</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">catastrophic forgetting</i>, which hinder their ability to maintain robust forward and backward learning capabilities. To address these issues, this paper proposes a novel Forward-Backward Knowledge Transfer (FBKT) paradigm, which strategically integrates forward distribution adaptation (FDA) and backward semantic alignment (BSA) mechanisms to achieve bidirectional adaptability in knowledge transfer. The FDA mechanism enhances forward adaptability by expanding and reserving the embedding space for new classes using semantic-irrelevant masked images as virtual negative classes, thereby mitigating data overfitting. It also employs self-supervised representation learning to utilize semantic-relevant local embeddings as additional positive samples, fostering class separation and generalization. Meanwhile, the BSA mechanism ensures the semantic consistency of previously learned classes across sessions during class-incremental learning, promoting smoother backward adaptability and reducing model degradation. Extensive experiments conducted on multiple benchmark datasets consistently highlight the superior performance and effectiveness of our FBKT compared to state-of-the-art methods.

3D face generation is a critical task for immersive multimedia applications, where a key challenge is the joint synthesis of expressive geometry and BRDF textures. Existing methods often struggle with geometric-textural coherence and corresponding reflectance modeling. To overcome these limitations, we present ExpDiff, a framework that generates expression meshes and corresponding BRDF textures from a single neutral-expression face. Our method employs an attention-based diffusion model to learn the semantic transition across expressions. To ensure correspondence between geometry and texture, we introduce a unified representation that explicitly models geometric-textural interaction, which is encoded into a shared latent space by models pre-trained on a vast dataset for strong generalization. To achieve semantically coherent and physically consistent generation, we propose to guide the denoising direction with specially designed textual prompts. We further construct two novel facial expression datasets, J-Reflectance, for ultra-high-quality assets, and FFHQ-BRDFExp for diverse identities, both of which are publicly released to advance the community. Extensive experiments demonstrate our method's superior performance in photo-realistic facial expression synthesis. Project page: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://cyh-sj.github.io/expdiff/</uri>.

Multi-Object Tracking (MOT) in Unmanned Aerial Vehicle (UAV) scenarios is characterized by frequent and abrupt camera motion, which presents two unique challenges: nonlinear motion and appearance degradation. Traditional motion models, designed for smooth and consistent motion, struggle to capture the complex background motion patterns caused by UAV movement; while appearance-based methods are easily disrupted by occlusion and blur, leading to unreliable associations. Even though dense optical flow is widely utilized to model complex motion patterns, the entanglement of background and object motion often introduces interference, limiting its effectiveness. To this end, we propose MoDe-Track, a unified framework that explicitly decouples scene motion into background and object components, and serves as an elegant integration of three robust components. Specifically, the Scene Motion Decomposition (SMD) module decouples the motion into the background and object components based on robust principal component analysis, serving as the foundation for motion compensation and feature propagation. Afterwards, the Background Motion Compensation (BMC) uses the decomposed background flow to estimate and compensate for camera motion, mitigating the effects of nonlinear motion. Finally, the Foreground-guided Feature Propagation (FFP) module uses the decoupled object flow to guide feature propagation across frames, achieving temporal consistency and enhancing robustness against occlusion and motion blur. Extensive experimental results on two benchmarks, VisDrone2019 and UAVDT, demonstrate that MoDe-Track consistently outperforms current multi-object tracking methods. We achieve 56.0% MOTA on VisDrone2019 and 56.2% MOTA on UAVDT, reaching the state-of-the-art among existing methods.

Artificial Intelligence Generated Content (AIGC) has grown rapidly in recent years, among which AI-based image generation has gained widespread attention due to its efficient and imaginative image creation ability. However, AI-generated Images (AIGIs) may not satisfy human preferences due to their unique distortions, which highlights the necessity to understand and evaluate human preferences for AIGIs. To this end, in this paper, we first establish a novel Image Quality Assessment (IQA) database for AIGIs, termed AIGCIQA2023+, which provides human visual preference scores and detailed preference explanations from three perspectives including quality, authenticity, and correspondence. Then, based on the constructed AIGCIQA2023+ database, this paper presents a <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MINT-IQA</b> model to evaluate and explain human preferences for AIGIs from <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</u>ulti-perspectives with <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IN</u>struction <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</u>uning. Specifically, the MINT-IQA model first learn and evaluate human preferences for AI-generated Images from multi-perspectives, then via the vision-language instruction tuning strategy, MINT-IQA attains powerful understanding and explanation ability for human visual preference on AIGIs, which can be used for feedback to further improve the assessment capabilities. Extensive experimental results demonstrate that the proposed MINT-IQA model achieves state-of-the-art performance in understanding and evaluating human visual preferences for AIGIs, and the proposed model also achieves competing results on traditional IQA tasks compared with state-of-the-art IQA models. The AIGCIQA2023+ database and MINT-IQA model are available at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/IntMeGroup/MINT-IQA</uri>

The challenges in single-image depth prediction (SIDP) are mainly due to the lack of smooth depth ground truth and the presence of irregular and complex objects. While window-based attention mechanisms, which balance long-range dependency capture with computational efficiency by processing elements within a fixed grid, have advanced SIDP research, they are limited by a constrained search range. This limitation can impede smooth depth estimation in irregularity and complexity. To address these challenges, we propose a novel attention mechanism that selectively identifies and aggregates only the most relevant information. Our approach enables flexible and efficient exploration by using data-dependent movable offsets to select substantial tokens and designating them as key-value pairs. Furthermore, we overcome the issue of small softmax values in traditional attention mechanisms through score-based grouping with top-k selection. Our feed-forward network, which incorporates a gating mechanism and grouped convolutions with varying cardinalities, refines features before passing them to subsequent layers, allowing for targeted focus on input features. Finally, we utilize feature maps from hierarchical decoders to estimate bin centers and per-pixel probability distributions. We introduce a 4-way selective scanning technique to aggregate these perpixel probability distributions smoothly, resulting in a dense and continuous depth map. The proposed network, named selective attention and selective aggregate depth (SA2Depth), demonstrates state-of-the-art performance across multiple datasets compared to previous methods. The code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/towardsDLCV/SA2DEPTH</uri>.