Alert button
Picture for Lei Luo

Lei Luo

Alert button

SGFeat: Salient Geometric Feature for Point Cloud Registration

Sep 12, 2023
Qianliang Wu, Yaqing Ding, Lei Luo, Chuanwei Zhou, Jin Xie, Jian Yang

Figure 1 for SGFeat: Salient Geometric Feature for Point Cloud Registration
Figure 2 for SGFeat: Salient Geometric Feature for Point Cloud Registration
Figure 3 for SGFeat: Salient Geometric Feature for Point Cloud Registration
Figure 4 for SGFeat: Salient Geometric Feature for Point Cloud Registration

Point Cloud Registration (PCR) is a critical and challenging task in computer vision. One of the primary difficulties in PCR is identifying salient and meaningful points that exhibit consistent semantic and geometric properties across different scans. Previous methods have encountered challenges with ambiguous matching due to the similarity among patch blocks throughout the entire point cloud and the lack of consideration for efficient global geometric consistency. To address these issues, we propose a new framework that includes several novel techniques. Firstly, we introduce a semantic-aware geometric encoder that combines object-level and patch-level semantic information. This encoder significantly improves registration recall by reducing ambiguity in patch-level superpoint matching. Additionally, we incorporate a prior knowledge approach that utilizes an intrinsic shape signature to identify salient points. This enables us to extract the most salient super points and meaningful dense points in the scene. Secondly, we introduce an innovative transformer that encodes High-Order (HO) geometric features. These features are crucial for identifying salient points within initial overlap regions while considering global high-order geometric consistency. To optimize this high-order transformer further, we introduce an anchor node selection strategy. By encoding inter-frame triangle or polyhedron consistency features based on these anchor nodes, we can effectively learn high-order geometric features of salient super points. These high-order features are then propagated to dense points and utilized by a Sinkhorn matching module to identify key correspondences for successful registration. In our experiments conducted on well-known datasets such as 3DMatch/3DLoMatch and KITTI, our approach has shown promising results, highlighting the effectiveness of our novel method.

Viaarxiv icon

Urban GeoBIM construction by integrating semantic LiDAR point clouds with as-designed BIM models

Apr 23, 2023
Jie Shao, Wei Yao, Puzuo Wang, Zhiyi He, Lei Luo

Figure 1 for Urban GeoBIM construction by integrating semantic LiDAR point clouds with as-designed BIM models
Figure 2 for Urban GeoBIM construction by integrating semantic LiDAR point clouds with as-designed BIM models
Figure 3 for Urban GeoBIM construction by integrating semantic LiDAR point clouds with as-designed BIM models
Figure 4 for Urban GeoBIM construction by integrating semantic LiDAR point clouds with as-designed BIM models

Developments in three-dimensional real worlds promote the integration of geoinformation and building information models (BIM) known as GeoBIM in urban construction. Light detection and ranging (LiDAR) integrated with global navigation satellite systems can provide geo-referenced spatial information. However, constructing detailed urban GeoBIM poses challenges in terms of LiDAR data quality. BIM models designed from software are rich in geometrical information but often lack accurate geo-referenced locations. In this paper, we propose a complementary strategy that integrates LiDAR point clouds with as-designed BIM models for reconstructing urban scenes. A state-of-the-art deep learning framework and graph theory are first combined for LiDAR point cloud segmentation. A coarse-to-fine matching program is then developed to integrate object point clouds with corresponding BIM models. Results show the overall segmentation accuracy of LiDAR datasets reaches up to 90%, and average positioning accuracies of BIM models are 0.023 m for pole-like objects and 0.156 m for buildings, demonstrating the effectiveness of the method in segmentation and matching processes. This work offers a practical solution for rapid and accurate urban GeoBIM construction.

Viaarxiv icon

AnyFlow: Arbitrary Scale Optical Flow with Implicit Neural Representation

Mar 29, 2023
Hyunyoung Jung, Zhuo Hui, Lei Luo, Haitao Yang, Feng Liu, Sungjoo Yoo, Rakesh Ranjan, Denis Demandolx

Figure 1 for AnyFlow: Arbitrary Scale Optical Flow with Implicit Neural Representation
Figure 2 for AnyFlow: Arbitrary Scale Optical Flow with Implicit Neural Representation
Figure 3 for AnyFlow: Arbitrary Scale Optical Flow with Implicit Neural Representation
Figure 4 for AnyFlow: Arbitrary Scale Optical Flow with Implicit Neural Representation

To apply optical flow in practice, it is often necessary to resize the input to smaller dimensions in order to reduce computational costs. However, downsizing inputs makes the estimation more challenging because objects and motion ranges become smaller. Even though recent approaches have demonstrated high-quality flow estimation, they tend to fail to accurately model small objects and precise boundaries when the input resolution is lowered, restricting their applicability to high-resolution inputs. In this paper, we introduce AnyFlow, a robust network that estimates accurate flow from images of various resolutions. By representing optical flow as a continuous coordinate-based representation, AnyFlow generates outputs at arbitrary scales from low-resolution inputs, demonstrating superior performance over prior works in capturing tiny objects with detail preservation on a wide range of scenes. We establish a new state-of-the-art performance of cross-dataset generalization on the KITTI dataset, while achieving comparable accuracy on the online benchmarks to other SOTA methods.

* CVPR 2023 (Highlight) 
Viaarxiv icon

Large-scale Point Cloud Registration Based on Graph Matching Optimization

Feb 16, 2023
Qianliang Wu, Yaqi Shen, Guofeng Mei, Yaqing Ding, Lei Luo, Jin Xie, Jian Yang

Figure 1 for Large-scale Point Cloud Registration Based on Graph Matching Optimization
Figure 2 for Large-scale Point Cloud Registration Based on Graph Matching Optimization
Figure 3 for Large-scale Point Cloud Registration Based on Graph Matching Optimization
Figure 4 for Large-scale Point Cloud Registration Based on Graph Matching Optimization

Point Clouds Registration is a fundamental and challenging problem in 3D computer vision. It has been shown that the isometric transformation is an essential property in rigid point cloud registration, but the existing methods only utilize it in the outlier rejection stage. In this paper, we emphasize that the isometric transformation is also important in the feature learning stage for improving registration quality. We propose a \underline{G}raph \underline{M}atching \underline{O}ptimization based \underline{Net}work (denoted as GMONet for short), which utilizes the graph matching method to explicitly exert the isometry preserving constraints in the point feature learning stage to improve %refine the point representation. Specifically, we %use exploit the partial graph matching constraint to enhance the overlap region detection abilities of super points ($i.e.,$ down-sampled key points) and full graph matching to refine the registration accuracy at the fine-level overlap region. Meanwhile, we leverage the mini-batch sampling to improve the efficiency of the full graph matching optimization. Given high discriminative point features in the evaluation stage, we utilize the RANSAC approach to estimate the transformation between the scanned pairs. The proposed method has been evaluated on the 3DMatch/3DLoMatch benchmarks and the KITTI benchmark. The experimental results show that our method achieves competitive performance compared with the existing state-of-the-art baselines.

Viaarxiv icon

Point Cloud Registration Based on Graph Matching Optimization

Feb 12, 2023
Qianliang Wu, Yaqi Shen, Guofeng Mei, Yaqing Ding, Lei Luo, Jin Xie, Jian Yang

Figure 1 for Point Cloud Registration Based on Graph Matching Optimization
Figure 2 for Point Cloud Registration Based on Graph Matching Optimization
Figure 3 for Point Cloud Registration Based on Graph Matching Optimization
Figure 4 for Point Cloud Registration Based on Graph Matching Optimization

Point Clouds Registration is a fundamental and challenging problem in 3D computer vision. It has been shown that the isometric transformation is an essential property in rigid point cloud registration, but the existing methods only utilize it in the outlier rejection stage. In this paper, we emphasize that the isometric transformation is also important in the feature learning stage for improving registration quality. We propose a \underline{G}raph \underline{M}atching \underline{O}ptimization based \underline{Net}work (denoted as GMONet for short), which utilizes the graph matching method to explicitly exert the isometry preserving constraints in the point feature learning stage to improve %refine the point representation. Specifically, we %use exploit the partial graph matching constraint to enhance the overlap region detection abilities of super points ($i.e.,$ down-sampled key points) and full graph matching to refine the registration accuracy at the fine-level overlap region. Meanwhile, we leverage the mini-batch sampling to improve the efficiency of the full graph matching optimization. Given high discriminative point features in the evaluation stage, we utilize the RANSAC approach to estimate the transformation between the scanned pairs. The proposed method has been evaluated on the 3DMatch/3DLoMatch benchmarks and the KITTI benchmark. The experimental results show that our method achieves competitive performance compared with the existing state-of-the-art baselines.

Viaarxiv icon

Curriculum Temperature for Knowledge Distillation

Dec 04, 2022
Zheng Li, Xiang Li, Lingfeng Yang, Borui Zhao, Renjie Song, Lei Luo, Jun Li, Jian Yang

Figure 1 for Curriculum Temperature for Knowledge Distillation
Figure 2 for Curriculum Temperature for Knowledge Distillation
Figure 3 for Curriculum Temperature for Knowledge Distillation
Figure 4 for Curriculum Temperature for Knowledge Distillation

Most existing distillation methods ignore the flexible role of the temperature in the loss function and fix it as a hyper-parameter that can be decided by an inefficient grid search. In general, the temperature controls the discrepancy between two distributions and can faithfully determine the difficulty level of the distillation task. Keeping a constant temperature, i.e., a fixed level of task difficulty, is usually sub-optimal for a growing student during its progressive learning stages. In this paper, we propose a simple curriculum-based technique, termed Curriculum Temperature for Knowledge Distillation (CTKD), which controls the task difficulty level during the student's learning career through a dynamic and learnable temperature. Specifically, following an easy-to-hard curriculum, we gradually increase the distillation loss w.r.t. the temperature, leading to increased distillation difficulty in an adversarial manner. As an easy-to-use plug-in technique, CTKD can be seamlessly integrated into existing knowledge distillation frameworks and brings general improvements at a negligible additional computation cost. Extensive experiments on CIFAR-100, ImageNet-2012, and MS-COCO demonstrate the effectiveness of our method. Our code is available at https://github.com/zhengli97/CTKD.

* AAAI 2023 
Viaarxiv icon

Denoising Multi-Similarity Formulation: A Self-paced Curriculum-Driven Approach for Robust Metric Learning

Dec 01, 2022
Chenkang Zhang, Lei Luo, Bin Gu

Figure 1 for Denoising Multi-Similarity Formulation: A Self-paced Curriculum-Driven Approach for Robust Metric Learning
Figure 2 for Denoising Multi-Similarity Formulation: A Self-paced Curriculum-Driven Approach for Robust Metric Learning
Figure 3 for Denoising Multi-Similarity Formulation: A Self-paced Curriculum-Driven Approach for Robust Metric Learning
Figure 4 for Denoising Multi-Similarity Formulation: A Self-paced Curriculum-Driven Approach for Robust Metric Learning

Deep Metric Learning (DML) is a group of techniques that aim to measure the similarity between objects through the neural network. Although the number of DML methods has rapidly increased in recent years, most previous studies cannot effectively handle noisy data, which commonly exists in practical applications and often leads to serious performance deterioration. To overcome this limitation, in this paper, we build a connection between noisy samples and hard samples in the framework of self-paced learning, and propose a \underline{B}alanced \underline{S}elf-\underline{P}aced \underline{M}etric \underline{L}earning (BSPML) algorithm with a denoising multi-similarity formulation, where noisy samples are treated as extremely hard samples and adaptively excluded from the model training by sample weighting. Especially, due to the pairwise relationship and a new balance regularization term, the sub-problem \emph{w.r.t.} sample weights is a nonconvex quadratic function. To efficiently solve this nonconvex quadratic problem, we propose a doubly stochastic projection coordinate gradient algorithm. Importantly, we theoretically prove the convergence not only for the doubly stochastic projection coordinate gradient algorithm, but also for our BSPML algorithm. Experimental results on several standard data sets demonstrate that our BSPML algorithm has better generalization ability and robustness than the state-of-the-art robust DML approaches.

Viaarxiv icon

Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report

Nov 07, 2022
Andrey Ignatov, Radu Timofte, Maurizio Denna, Abdel Younes, Ganzorig Gankhuyag, Jingang Huh, Myeong Kyun Kim, Kihwan Yoon, Hyeon-Cheol Moon, Seungho Lee, Yoonsik Choe, Jinwoo Jeong, Sungjei Kim, Maciej Smyl, Tomasz Latkowski, Pawel Kubik, Michal Sokolski, Yujie Ma, Jiahao Chao, Zhou Zhou, Hongfan Gao, Zhengfeng Yang, Zhenbing Zeng, Zhengyang Zhuge, Chenghua Li, Dan Zhu, Mengdi Sun, Ran Duan, Yan Gao, Lingshun Kong, Long Sun, Xiang Li, Xingdong Zhang, Jiawei Zhang, Yaqi Wu, Jinshan Pan, Gaocheng Yu, Jin Zhang, Feng Zhang, Zhe Ma, Hongbin Wang, Hojin Cho, Steve Kim, Huaen Li, Yanbo Ma, Ziwei Luo, Youwei Li, Lei Yu, Zhihong Wen, Qi Wu, Haoqiang Fan, Shuaicheng Liu, Lize Zhang, Zhikai Zong, Jeremy Kwon, Junxi Zhang, Mengyuan Li, Nianxiang Fu, Guanchen Ding, Han Zhu, Zhenzhong Chen, Gen Li, Yuanfan Zhang, Lei Sun, Dafeng Zhang, Neo Yang, Fitz Liu, Jerry Zhao, Mustafa Ayazoglu, Bahri Batuhan Bilecen, Shota Hirose, Kasidis Arunruangsirilert, Luo Ao, Ho Chun Leung, Andrew Wei, Jie Liu, Qiang Liu, Dahai Yu, Ao Li, Lei Luo, Ce Zhu, Seongmin Hong, Dongwon Park, Joonhee Lee, Byeong Hyun Lee, Seunggyu Lee, Se Young Chun, Ruiyuan He, Xuhao Jiang, Haihang Ruan, Xinjian Zhang, Jing Liu, Garas Gendy, Nabil Sabor, Jingchao Hou, Guanghui He

Figure 1 for Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report
Figure 2 for Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report
Figure 3 for Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report
Figure 4 for Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report

Image super-resolution is a common task on mobile and IoT devices, where one often needs to upscale and enhance low-resolution images and video frames. While numerous solutions have been proposed for this problem in the past, they are usually not compatible with low-power mobile NPUs having many computational and memory constraints. In this Mobile AI challenge, we address this problem and propose the participants to design an efficient quantized image super-resolution solution that can demonstrate a real-time performance on mobile NPUs. The participants were provided with the DIV2K dataset and trained INT8 models to do a high-quality 3X image upscaling. The runtime of all models was evaluated on the Synaptics VS680 Smart Home board with a dedicated edge NPU capable of accelerating quantized neural networks. All proposed solutions are fully compatible with the above NPU, demonstrating an up to 60 FPS rate when reconstructing Full HD resolution images. A detailed description of all models developed in the challenge is provided in this paper.

* arXiv admin note: text overlap with arXiv:2105.07825, arXiv:2105.08826, arXiv:2211.04470, arXiv:2211.03885, arXiv:2211.05256 
Viaarxiv icon

Sampling Through the Lens of Sequential Decision Making

Aug 17, 2022
Jason Xiaotian Dou, Alvin Qingkai Pan, Runxue Bao, Haiyi Harry Mao, Lei Luo

Figure 1 for Sampling Through the Lens of Sequential Decision Making
Figure 2 for Sampling Through the Lens of Sequential Decision Making
Figure 3 for Sampling Through the Lens of Sequential Decision Making
Figure 4 for Sampling Through the Lens of Sequential Decision Making

Sampling is ubiquitous in machine learning methodologies. Due to the growth of large datasets and model complexity, we want to learn and adapt the sampling process while training a representation. Towards achieving this grand goal, a variety of sampling techniques have been proposed. However, most of them either use a fixed sampling scheme or adjust the sampling scheme based on simple heuristics. They cannot choose the best sample for model training in different stages. Inspired by "Think, Fast and Slow" (System 1 and System 2) in cognitive science, we propose a reward-guided sampling strategy called Adaptive Sample with Reward (ASR) to tackle this challenge. To the best of our knowledge, this is the first work utilizing reinforcement learning (RL) to address the sampling problem in representation learning. Our approach optimally adjusts the sampling process to achieve optimal performance. We explore geographical relationships among samples by distance-based sampling to maximize overall cumulative reward. We apply ASR to the long-standing sampling problems in similarity-based loss functions. Empirical results in information retrieval and clustering demonstrate ASR's superb performance across different datasets. We also discuss an engrossing phenomenon which we name as "ASR gravity well" in experiments.

Viaarxiv icon