检索结果-内蒙古大学图书馆

GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection 39

学校读者我要写书评

暂无评论

GeoBEV: Learning Geometric BEV Representation for Multi-view...

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Zhang, Jinqing Zhang, Yanan Qi, Yunlong Fu, Zehua Liu, Qingjie Wang, Yunhong State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing China Beijing Jingwei Hirain Technologies Co. Inc. China Hangzhou Innovation Institute Beihang University Hangzhou China Zhongguancun Laboratory Beijing China

ISBN: (纸本)157735897X

Bird’s-Eye-View (BEV) representation has emerged as a mainstream paradigm for multi-view 3D object detection, demonstrating impressive perceptual capabilities. However, existing methods overlook the geometric quality of BEV representation, leaving it in a low-resolution state and failing to restore the authentic geometric information of the scene. In this paper, we identify the drawbacks of previous approaches that limit the geometric quality of BEV representation and propose Radial-Cartesian BEV Sampling (RC-Sampling), which outperforms other feature transformation methods in efficiently generating high-resolution dense BEV representation to restore fine-grained geometric information. Additionally, we design a novel In-Box Label to substitute the traditional depth label generated from the LiDAR points. This label reflects the actual geometric structure of objects rather than just their surfaces, injecting real-world geometric information into the BEV representation. In conjunction with the In-Box Label, Centroid-Aware Inner Loss (CAI Loss) is developed to capture the inner geometric structure of objects. Finally, we integrate the aforementioned modules into a novel multi-view 3D object detector, dubbed GeoBEV, which achieves a state-of-the-art result of 66.2% NDS on the nuScenes test set. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Geometry

PACF: Prototype Augmented Compact Features for Improving Domain Adaptive Object Detection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Liu, Chenguang Feng, Yongchao Zhang, Yanan Liu, Qingjie Wang, Yunhong State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing100191 China School of Computer Science and Information Engineering Hefei University of Technology Hefei230601 China Hangzhou Innovation Institute Beihang University Hangzhou310051 China

In recent years, there has been significant advancement in object detection. However, applying off-the-shelf detectors to a new domain leads to significant performance drop, caused by the domain gap. These detectors exhibit higher-variance class-conditional distributions in the target domain than that in the source domain, along with mean shift. To address this problem, we propose the Prototype Augmented Compact Features (PACF) framework to regularize the distribution of intra-class features. Specifically, we provide an in-depth theoretical analysis on the lower bound of the target features-related likelihood and derive the prototype cross entropy loss to further calibrate the distribution of target RoI features. Furthermore, a mutual regularization strategy is designed to enable the linear and prototype-based classifiers to learn from each other, promoting feature compactness while enhancing discriminability. Thanks to this PACF framework, we have obtained a more compact cross-domain feature space, within which the variance of the target features’ class-conditional distributions has significantly decreased, and the class-mean shift between the two domains has also been further reduced. The results on different adaptation settings are state-of-the-art, which demonstrate the board applicability and effectiveness of the proposed approach. Copyright © 2025, The Authors. All rights reserved.

关键词： Object detection

Scaffold-BPE: Enhancing Byte Pair Encoding for Large Language Models with Simple and Effective Scaffold Token Removal 39

学校读者我要写书评

暂无评论

Scaffold-BPE: Enhancing Byte Pair Encoding for Large Languag...

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Lian, Haoran Xiong, Yizhe Niu, Jianwei Mo, Shasha Su, Zhenpeng Lin, Zijia Chen, Hui Han, Jungong Ding, Guiguang Beihang University China Tsinghua University China BNRist China State Key Laboratory of Virtual Reality Technology and Systems Beihang University China Zhongguancun Laboratory China Zhengzhou University Research Institute of Industrial Technology Zhengzhou University China Chinese Academy of Sciences China

ISBN: (纸本)157735897X

Byte Pair Encoding (BPE) serves as a foundation method for text tokenization in the Natural Language Processing (NLP) field. Despite its wide adoption, the original BPE algorithm harbors an inherent flaw: it inadvertently introduces a frequency imbalance for tokens in the text corpus. Since BPE iteratively merges the most frequent token pair in the text corpus to generate a new token and keeps all generated tokens in the vocabulary, it unavoidably holds tokens that primarily act as components of a longer token and appear infrequently on their own. We term such tokens as Scaffold Tokens. Due to their infrequent occurrences in the text corpus, Scaffold Tokens pose a learning imbalance issue. To address that issue, we propose Scaffold-BPE, which incorporates a dynamic scaffold token removal mechanism by parameter-free, computation-light, and easy-to-implement modifications to the original BPE method. This novel approach ensures the exclusion of low-frequency Scaffold Tokens from the token representations for given texts, thereby mitigating the issue of frequency imbalance and facilitating model training. On extensive experiments across language modeling and even machine translation, Scaffold-BPE consistently outperforms the original BPE, well demonstrating its effectiveness. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Scaffolds (biology)

YOLC: You Only Look Clusters for Tiny Object Detection in Aerial Images

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liu, Chenguang Gao, Guangshuai Huang, Ziyue Hu, Zhenghui Liu, Qingjie Wang, Yunhong The State Key Laboratory of Virtual Reality Technology and Systems Beihang University Haidian Beijing100191 China The Hangzhou Innovation Institute Beihang University Hangzhou310051 China The School of Electronics and Information Zhongyuan University of Technology Zhengzhou450007 China

Detecting objects from aerial images poses significant challenges due to the following factors: 1) Aerial images typically have very large sizes, generally with millions or even hundreds of millions of pixels, while computational resources are limited. 2) Small object size leads to insufficient information for effective detection. 3) Non-uniform object distribution leads to computational resource wastage. To address these issues, we propose YOLC (You Only Look Clusters), an efficient and effective framework that builds on an anchor-free object detector, CenterNet. To overcome the challenges posed by large-scale images and non-uniform object distribution, we introduce a Local Scale Module (LSM) that adaptively searches cluster regions for zooming in for accurate detection. Additionally, we modify the regression loss using Gaussian Wasserstein distance (GWD) to obtain high-quality bounding boxes. Deformable convolution and refinement methods are employed in the detection head to enhance the detection of small objects. We perform extensive experiments on two aerial image datasets, including Visdrone2019 and UAVDT, to demonstrate the effectiveness and superiority of our proposed approach. Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

One-Shot Neural Band Selection for Spectral Recovery

学校读者我要写书评

暂无评论

One-Shot Neural Band Selection for Spectral Recovery

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Hai-Miao Hu Zhenbo Xu Wenshuai Xu You Song YiTao Zhang Liu Liu Zhilin Han Ajin Meng Hangzhou Innovation Institute Beihang University Hangzhou China State Key Laboratory of Virtual Reality Technology and Systems Beihang University School of Software Beihang University ShiFang Technology Inc Hangzhou China

Band selection has a great impact on the spectral recovery quality. To solve this ill-posed inverse problem, most band selection methods adopt hand-crafted priors or exploit clustering or sparse regularization constraints to find most prominent bands. These methods are either very slow due to the computational cost of repeatedly training with respect to different selection frequencies or different band combinations. Many traditional methods rely on the scene prior and thus are not applicable to other scenarios. In this paper, we present a novel one-shot Neural Band Selection (NBS) framework for spectral recovery. Unlike conventional searching approaches with a discrete search space and a non-differentiable search strategy, our NBS is based on the continuous relaxation of the band selection process, thus allowing efficient band search using gradient descent. To enable the compatibility for selecting any number of bands in one-shot, we further exploit the band-wise correlation matrices to progressively suppress similar adjacent bands. Extensive evaluations on the NTIRE 2022 Spectral Reconstruction Challenge demonstrate that our NBS achieves consistent performance gains over competitive baselines when examined with four different spectral recovery methods. Our code will be publicly available.

关键词： Training Correlation Inverse problems NIST Signal processing Performance gain Search problems

Disentangling Local and Global Information for Light Field Depth Estimation

学校读者我要写书评

暂无评论

Disentangling Local and Global Information for Light Field D...

2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2023

作者： Yang, Xueting Deng, Junli Chen, Rongshan Cong, Ruixuan Ke, Wei Sheng, Hao Communication University of China School of Information and Communication Engineering Beijing100024 China Beihang University State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beijing100191 China Beihang Hangzhou Innovation Institute Yuhang Xixi Octagon City Hangzhou310023 China Macao Polytechnic University Faculty of Applied Sciences 999078 China

ISBN: (纸本)9798350302493

Accurate depth estimation from light field images is essential for various applications. Deep learning-based techniques have shown great potential in addressing this problem while still face challenges such as sensitivity to occlusions and difficulties in handling untextured areas. To overcome these limitations, we propose a novel approach that utilizes both local and global features in the cost volume for depth estimation. Specifically, our hybrid cost volume network consists of two complementary sub-modules: a 2D ContextNet for global context information and a matching cost volume for local feature information. We also introduce an occlusion-aware loss that accounts for occlusion areas to improve depth estimation quality. We demonstrate the effectiveness of our approach on the UrbanLF and HCInew datasets, showing significant improvements over existing methods, especially in occluded and untextured regions. Our method disentangles local feature and global semantic information explicitly, reducing the occlusion and untextured area reconstruction error and improving the accuracy of depth estimation. © 2023 IEEE.

关键词： Semantics

Generic Knowledge Boosted Pre-training For Remote Sensing Images

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Huang, Ziyue Zhang, Mingming Gong, Yuan Liu, Qingjie Wang, Yunhong State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing100191 China Hangzhou Innovation Institute Beihang University Hangzhou310051 China School of Software and Microelectronics Peking University Beijing100871 China

—Deep learning models are essential for scene classification, change detection, land cover segmentation, and other remote sensing image understanding tasks. Most backbones of existing remote sensing deep learning models are typically initialized by pre-trained weights obtained from ImageNet pretraining (IMP). However, domain gaps exist between remote sensing images and natural images (e.g., ImageNet), making deep learning models initialized by pre-trained weights of IMP perform poorly for remote sensing image understanding. Although some pre-training methods are studied in the remote sensing community, current remote sensing pre-training methods face the problem of vague generalization by only using remote sensing images. In this paper, we propose a novel remote sensing pre-training framework, Generic Knowledge Boosted Remote Sensing Pre-training (GeRSP), to learn robust representations from remote sensing and natural images for remote sensing understanding tasks. GeRSP contains two pre-training branches: (1) A self-supervised pre-training branch is adopted to learn domain-related representations from unlabeled remote sensing images. (2) A supervised pre-training branch is integrated into GeRSP for general knowledge learning from labeled natural images. Moreover, GeRSP combines two pre-training branches using a teacher-student architecture to simultaneously learn representations with general and special knowledge, which generates a powerful pre-trained model for deep learning model initialization. Finally, we evaluate GeRSP and other remote sensing pre-training methods on three downstream tasks, i.e., object detection, semantic segmentation, and scene classification. The extensive experimental results consistently demonstrate that GeRSP can effectively learn robust representations in a unified manner, improving the performance of remote sensing downstream tasks. Copyright © 2024, The Authors. All rights reserved.

关键词： Remote sensing

GeoBEV: Learning Geometric BEV Representation for Multi-view 3D Object Detection

学校读者我要写书评

暂无评论

arXiv 2024年

Bird’s-Eye-View (BEV) representation has emerged as a mainstream paradigm for multi-view 3D object detection, demonstrating impressive perceptual capabilities. However, existing methods overlook the geometric quality of BEV representation, leaving it in a low-resolution state and failing to restore the authentic geometric information of the scene. In this paper, we identify the drawbacks of previous approaches that limit the geometric quality of BEV representation and propose Radial-Cartesian BEV Sampling (RC-Sampling), which outperforms other feature transformation methods in efficiently generating high-resolution dense BEV representation to restore fine-grained geometric information. Additionally, we design a novel In-Box Label to substitute the traditional depth label generated from the LiDAR points. This label reflects the actual geometric structure of objects rather than just their surfaces, injecting real-world geometric information into the BEV representation. In conjunction with the In-Box Label, Centroid-Aware Inner Loss (CAI Loss) is developed to capture the inner geometric structure of objects. Finally, we integrate the aforementioned modules into a novel multi-view 3D object detector, dubbed GeoBEV, which achieves a state-of-the-art result of 66.2% NDS on the nuScenes test set. The code is available at https://***/mengtan00/***. Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

UNeLF: Unconstrained Neural Light Field for Self-Supervised Angular Super-Resolution

学校读者我要写书评

暂无评论

IEEE Transactions on Circuits and systems for Video technology 2025年

作者： Zhao, Mingyuan Sheng, Hao Chen, Rongshan Cong, Ruixuan Cui, Zhenglong Yang, Da Beihang University State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beijing100191 China Beihang University Data Science and Intelligent Computing Laboratory Hangzhou International Innovation Institute Hangzhou311115 China Macao Polytechnic University Faculty of Applied Sciences 999078 China

Compared to supervised learning methods, self-supervised learning methods address the domain gap problem between light field (LF) datasets collected under varying acquisition conditions, which typically leads to decreased performance when differences exist in the distribution between the training and test sets. However, current self-supervised light field angular super-resolution (LFASR) techniques primarily focus on exploiting discrete spatial-angular features while neglecting continuous LF information. In contrast to previous work, we propose a self-supervised unconstrained neural light field (UNeLF) to continuously represent LF for LFASR. Specifically, any LF can be described as the camera pose for each sub-aperture image (SAI) and the two-plane that captures these SAIs. To describe the former, we introduce a SAIs-dependent pose optimization method to solve the issue that arises from the narrow baseline of most LF data, which hinders robust camera pose estimation. This mechanism reduces the number of trainable camera parameters from a quadratic to a constant scale, thereby alleviating the complexity of joint optimization. For the latter, we propose a novel adaptive two-plane parameterization strategy to determine the two-plane that captures these SAIs, facilitating refocusing. Finally, we jointly optimize the camera parameters, near-far planes and neural light field, efficiently mapping each adaptive two-plane parameterized ray to its correspondence color in a continuous manner. Comprehensive experiments demonstrate that UNeLF achieves faster training and inference with fewer computational resources while exhibiting superior performance on both synthetic and real-world datasets. © 1991-2012 IEEE.

关键词： Self-supervised learning