Recent advances in long video understanding typically mitigate visual redundancy through visual token pruning based on attention distribution. However, while existing methods employ post-hoc low-response token pruning in decoder layers, they overlook the input-level semantic correlation between visual tokens and instructions (query). In this paper, we propose QuoTA, an ante-hoc training-free modular that extends existing large video-language models (LVLMs) for visual token assignment based on query-oriented frame-level importance assessment. The query-oriented token selection is crucial as it aligns visual processing with task-specific requirements, optimizing token budget utilization while preserving semantically relevant content. Specifically, (i) QuoTA strategically allocates frame-level importance scores based on query relevance, enabling one-time visual token assignment before cross-modal interactions in decoder layers, (ii) we decouple the query through Chain-of-Thoughts reasoning to facilitate more precise LVLM-based frame importance scoring, and (iii) QuoTA offers a plug-and-play functionality that extends to existing LVLMs. Extensive experimental results demonstrate that implementing QuoTA with LLaVA-Video-7B yields an average performance improvement of 3.2% across six benchmarks (including Video-MME and MLVU) while operating within an identical visual token budget as the baseline.

Recent Multimodal Large Language Models (MLLMs) have typically focused on integrating visual and textual modalities, with less emphasis placed on the role of speech in enhancing interaction. However, speech plays a crucial role in multimodal dialogue systems, and implementing high-performance in both vision and speech tasks remains a challenge due to the fundamental modality differences. In this paper, we propose a carefully designed multi-stage training methodology that progressively trains LLM to understand both visual and speech information, ultimately enabling fluent vision and speech interaction. Our approach not only preserves strong vision-language capacity, but also enables efficient speech-to-speech dialogue capabilities without separate ASR and TTS modules, significantly accelerating multimodal end-to-end response speed. By comparing against state-of-the-art counterparts across benchmarks for image, video, and speech, we demonstrate that our omni model is equipped with both strong visual and speech capabilities, making omni understanding and interaction.

Existing large video-language models (LVLMs) struggle to comprehend long videos correctly due to limited context. To address this problem, fine-tuning long-context LVLMs and employing GPT-based agents have emerged as promising solutions. However, fine-tuning LVLMs would require extensive high-quality data and substantial GPU resources, while GPT-based agents would rely on proprietary models (e.g., GPT-4o). In this paper, we propose Video Retrieval-Augmented Generation (Video-RAG), a training-free and cost-effective pipeline that employs visually-aligned auxiliary texts to help facilitate cross-modality alignment while providing additional information beyond the visual content. Specifically, we leverage open-source external tools to extract visually-aligned information from pure video data (e.g., audio, optical character, and object detection), and incorporate the extracted information into an existing LVLM as auxiliary texts, alongside video frames and queries, in a plug-and-play manner. Our Video-RAG offers several key advantages: (i) lightweight with low computing overhead due to single-turn retrieval; (ii) easy implementation and compatibility with any LVLM; and (iii) significant, consistent performance gains across long video understanding benchmarks, including Video-MME, MLVU, and LongVideoBench. Notably, our model demonstrates superior performance over proprietary models like Gemini-1.5-Pro and GPT-4o when utilized with a 72B model.

Recent advances in 3D point cloud analysis bring a diverse set of network architectures to the field. However, the lack of a unified framework to interpret those networks makes any systematic comparison, contrast, or analysis challenging, and practically limits healthy development of the field. In this paper, we take the initiative to explore and propose a unified framework called PointMeta, to which the popular 3D point cloud analysis approaches could fit. This brings three benefits. First, it allows us to compare different approaches in a fair manner, and use quick experiments to verify any empirical observations or assumptions summarized from the comparison. Second, the big picture brought by PointMeta enables us to think across different components, and revisit common beliefs and key design decisions made by the popular approaches. Third, based on the learnings from the previous two analyses, by doing simple tweaks on the existing approaches, we are able to derive a basic building block, termed PointMetaBase. It shows very strong performance in efficiency and effectiveness through extensive experiments on challenging benchmarks, and thus verifies the necessity and benefits of high-level interpretation, contrast, and comparison like PointMeta. In particular, PointMetaBase surpasses the previous state-of-the-art method by 0.7%/1.4/%2.1% mIoU with only 2%/11%/13% of the computation cost on the S3DIS datasets. The code and models are available at https://github.com/linhaojia13/PointMetaBase.

Recently, it has been shown that the traditional Image Signal Processing (ISP) can be replaced by deep neural networks due to their superior performance. However, most of these networks require heavy computation burden and thus are far from sufficient to be deployed on resource-limited platforms, including but not limited to mobile devices and FPGA. To tackle this challenge, we propose an automated search framework that derives ISP models with high image quality while satisfying the low-computation requirement. To reduce the search cost, we adopt the weight-sharing strategy by introducing a supernet and decouple the architecture search into two stages, supernet training and hard-aware evolutionary search. With the proposed framework, we can train the ISP model once and quickly find high-performance but low-computation models on multiple devices. Experiments demonstrate that the searched ISP models have an excellent trade-off between image quality and model complexity, i.e., achieve compelling reconstruction quality with more than 90% reduction in FLOPs as compared to the state-of-the-art networks.

Deep learning for 3-D point cloud perception has been a very active research topic in recent years. A current trend is toward the combination of the semantically strong and the fine-grained information from different scales of intermediate representations to boost network generalization power and robustness against scale variation. One prominent challenge is how to effectively conduct the allocation of multiple scales of information. In this letter, we propose a module, named adaptive pyramid context fusion (APCF), to adaptively capture scales of contextual information from a multiscale feature pyramid for the point cloud. The APCF module reweights and aggregates the features from different levels in the feature pyramid via a softmax attention strategy. The allocation of information is adaptively conducted level by level from bottom to up first and then from top to bottom. To ensure both effectiveness and efficiency, we propose a multiscale context-aware network APCF-Net through applying our proposed APCF to the PointConv architecture. Experiments demonstrate that APCF-Net surpasses its vanilla counterpart by a large margin both in effectiveness and efficiency. Especially, APCF-Net outperforms state-of-the-art approaches on 3-D object classification and semantic segmentation task, with the overall accuracy of 93.3% on ModelNet40 and mIoU of 63.1% on ScanNet V2 online test.

As a foundational preprocessing step for a lot of downstream tasks, ground filtering from airborne LiDAR data is designed to separate the ground points and preserve the off-ground points with complete shape information. However, because of the undulating terrain, it is still a challenge work to filter the ground under complex mountain regions. In this paper, we provide a deep learning based model to improve the ground filtering performance in abrupt slope using airborne LiDAR point clouds. Specifically, we first design a local topological information mining module to extract the local features. Then a modified graph convolutional networks (GCNs) is developed to fusion the local features and global features. Compared with most existing methods, our model not only enjoys the parameter-free advantage, which means it can be applied easily in various areas, but also obtains better ground filtering performance and can preserve more complete information contained in off-ground points. Experiments was implemented on seven forest areas. The proposed method obtains promising ground filtering results with mean total error of 6.46% and the mean kappa coefficient of 86.01%.