Current multimodal large language models (MLLMs) struggle with hour-level video understanding, facing significant challenges not only in modeling the substantial information volume of long videos but also in overcoming the memory wall and resource constraints during both training and inference. Although recent training-free approaches have alleviated resource demands by compressing visual features, their reliance on incomplete visual information limits the performance potential. To address these limitations, we propose Adaptive Pivot Visual information Retrieval (APVR), a training-free framework that hierarchically retrieves and retains sufficient and important visual information. It breakthroughs the memory wall limitation via two complementary components: Pivot Frame Retrieval employs query expansion and iterative spatio-semantic confidence scoring to identify relevant video frames, and Pivot Token Retrieval performs query-aware attention-driven token selection within up to 1024 pivot frames. This dual granularity approach enables the processing of hour-long videos while maintaining semantic fidelity. Experimental validations on three different baseline MLLMs demonstrate significant performance improvements up to 9.5%, 4.6% and 9.7% on LongVideoBench, VideoMME and MLVU, respectively. APVR achieves state-of-the-art results for both training-free and training-based approaches.

The 4D millimeter wave radar has gained increasing attention in autonomous driving due to its robustness against environmental interference compared to other perception devices such as cameras or LiDAR. However, the practical deployment of radar remains challenging due to the noise and high sparsity of radar point clouds, as well as the resource limitations of edge computing. To address these issues, we propose Radar-Mamba, a lightweight and efficient radar enhancement approach for boosting radar perception. The proposed approach contains three main components: cross-modal alignment, radar enhancement architecture based on the Mamba model, and Doppler feature fusion. Specifically, the point clouds of radar and LiDAR are first refined and aligned to build more dense and rich occupancies. The aligned 4D radar data is then enhanced by capturing both local and global spatial-temporal features, while integrating radar-specific velocity and elevation information for further denoising. Experiments on two open-source datasets demonstrate that our method achieves state-of-the-art performance and generates high-quality 4D point clouds with a density comparable to LiDAR while maintaining a low parameter count friendly for practical deployment.

As a special subset of multi-view settings for 3D human pose estimation, stereoscopic settings show promising applications in practice since they are not ill-posed but could be as mobile as monocular ones. However, when there are only two views, the problems of occlusions and “double counting” (ambiguity between symmetric joints) pose greater challenges that are not addressed by previous approaches. On this concern, we propose a novel framework to detect limb orientations in field form and incorporate them explicitly with joint features. Two modules are proposed to realize the fusion. At 3D level, we designcompound triangulationas an explicit module that produces the optimal pose using 2D joint locations and limb orientations. The module is derived from reformulating triangulation in 3D space, and expanding it with the optimization of limb orientations. At 2D level, we propose a parameter-free module namedco-fixingto enable joint and limb features to fix each other to alleviate the impact of “double counting.” Features from both parts are first used to infer each other via simple convolutions and then fixed by the inferred ones respectively. We test our method on two public benchmarks, Human3.6M and Total Capture, and our method achieves state-of-the-art performance on stereoscopic settings and comparable results on common 4-view benchmarks.

Haplotype networks are graphs used to represent evolutionary relationships between a set of taxa and are characterized by intuitiveness in analyzing genealogical relationships of closely related genomes. We here propose a novel algorithm termed McAN that considers mutation spectrum history (mutations in ancestry haplotype should be contained in descendant haplotype), node size (corresponding to sample count for a given node) and sampling time when constructing haplotype network. We show that McAN is two orders of magnitude faster than state-of-the-art algorithms without losing accuracy, making it suitable for analysis of a large number of sequences. Based on our algorithm, we developed an online web server and offline tool for haplotype network construction, community lineage determination, and interactive network visualization. We demonstrate that McAN is highly suitable for analyzing and visualizing massive genomic data and is helpful to enhance the understanding of genome evolution. Availability: Source code is written in C/C++ and available at https://github.com/Theory-Lun/McAN and https://ngdc.cncb.ac.cn/biocode/tools/BT007301 under the MIT license. Web server is available at https://ngdc.cncb.ac.cn/bit/hapnet/. SARS-CoV-2 dataset are available at https://ngdc.cncb.ac.cn/ncov/. Contact: [email protected] (Song S), [email protected] (Zhao W), [email protected] (Bao Y), [email protected] (Zhang Z), [email protected] (Xue Y).

Commercial motion-capture systems produce excell- ent in-studio reconstructions, but offer no comparable solution for acquisition in everyday environments. We present a system for acquiring motions almost anywhere. This wearable system gathers ultrasonic time-of-flight and inertial measurements with a set of inexpensive miniature sensors worn on the garment. After recording, the information is combined using an Extended Kalman Filter to reconstruct joint configurations of a body. Experimental results show that even motions that are traditionally difficult to acquire are recorded with ease within their natural settings. Although our prototype does not reliably recover the global transformation, we show that the resulting motions are visually similar to the original ones, and that the combined acoustic and intertial system reduces the drift commonly observed in purely inertial systems. Our final results suggest that this system could become a versatile input device for a variety of augmented-reality applications.

We propose a conceptually simple framework for fast COVID-19 screening in 3D chest CT images. The framework can efficiently predict whether or not a CT scan contains pneumonia while simultaneously identifying pneumonia types between COVID-19 and Interstitial Lung Disease (ILD) caused by other viruses. In the proposed method, two 3D-ResNets are coupled together into a single model for the two above-mentioned tasks via a novel prior-attention strategy. We extend residual learning with the proposed prior-attention mechanism and design a new so-called prior-attention residual learning (PARL) block. The model can be easily built by stacking the PARL blocks and trained end-to-end using multi-task losses. More specifically, one 3D-ResNet branch is trained as a binary classifier using lung images with and without pneumonia so that it can highlight the lesion areas within the lungs. Simultaneously, inside the PARL blocks, prior-attention maps are generated from this branch and used to guide another branch to learn more discriminative representations for the pneumonia-type classification. Experimental results demonstrate that the proposed framework can significantly improve the performance of COVID-19 screening. Compared to other methods, it achieves a state-of-the-art result. Moreover, the proposed method can be easily extended to other similar clinical applications such as computer-aided detection and diagnosis of pulmonary nodules in CT images, glaucoma lesions in Retina fundus images, etc.

MOTIVATION: The significance of long non-coding RNAs (lncRNAs) in many biological processes and diseases has gained intense interests over the past several years. However, computational identification of lncRNAs in a wide range of species remains challenging; it requires prior knowledge of well-established sequences and annotations or species-specific training data, but the reality is that only a limited number of species have high-quality sequences and annotations. RESULTS: Here we first characterize lncRNAs in contrast to protein-coding RNAs based on feature relationship and find that the feature relationship between open reading frame length and guanine-cytosine (GC) content presents universally substantial divergence in lncRNAs and protein-coding RNAs, as observed in a broad variety of species. Based on the feature relationship, accordingly, we further present LGC, a novel algorithm for identifying lncRNAs that is able to accurately distinguish lncRNAs from protein-coding RNAs in a cross-species manner without any prior knowledge. As validated on large-scale empirical datasets, comparative results show that LGC outperforms existing algorithms by achieving higher accuracy, well-balanced sensitivity and specificity, and is robustly effective (>90% accuracy) in discriminating lncRNAs from protein-coding RNAs across diverse species that range from plants to mammals. To our knowledge, this study, for the first time, differentially characterizes lncRNAs and protein-coding RNAs based on feature relationship, which is further applied in computational identification of lncRNAs. Taken together, our study represents a significant advance in characterization and identification of lncRNAs and LGC thus bears broad potential utility for computational analysis of lncRNAs in a wide range of species. AVAILABILITY AND IMPLEMENTATION: LGC web server is publicly available at http://bigd.big.ac.cn/lgc/calculator. The scripts and data can be downloaded at http://bigd.big.ac.cn/biocode/tools/BT000004. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: With the amount of influenza genome sequence data growing rapidly, researchers need machine assistance in selecting datasets and exploring the data. Enhanced visualization tools are required to represent results of the exploratory analysis on the web in an easy-to-comprehend form and to facilitate convenient information retrieval. RESULTS: We developed an approach to visualize large phylogenetic trees in an aggregated form with a special representation of subscale details. The initial aggregated tree representation is built with a level of resolution automatically selected to fit into the available screen space, with terminal groups selected based on sequence similarity. The default aggregated representation can be refined by users interactively.Structure and data variability within terminal groups are displayed using small trees that have the same vertical size as the text annotation of the group. These subscale representations are calculated using systematic sampling from the corresponding terminal group. The aggregated tree containing terminal groups can be annotated using aggregation of structured metadata, such as seasonal distribution, geographic locations, etc. AVAILABILITY: The algorithms are implemented in JavaScript within the NCBI Influenza Virus Resource 1.