Abstract
ROS 2 is rapidly becoming a standard in the robotics industry. Built upon DDS as its default communication middleware and used in safety-critical scenarios, adding security to robots and ROS computational graphs is increasingly becoming a concern. The present work introduces SROS2, a series of developer tools and libraries that facilitate adding security to ROS 2 graphs. Focusing on a usability-centric approach in SROS2, we present a methodology for securing graphs systematically while following the DevSecOps model. We also demonstrate the use of our security tools by presenting an application case study that considers securing a graph using the popular Navigation2 and SLAM Toolbox stacks applied in a TurtleBot3 robot. We analyse the current capabilities of SROS2 and discuss the shortcomings, which provides insights for future contributions and extensions. Ultimately, we present SROS2 as usable security tools for ROS 2 and argue that without usability, security in robotics will be greatly impaired.
Keyword: odometry
There is no result
Keyword: livox
There is no result
Keyword: loam
There is no result
Keyword: lidar
Estimating Visual Information From Audio Through Manifold Learning
Abstract
We propose a new framework for extracting visual information about a scene only using audio signals. Audio-based methods can overcome some of the limitations of vision-based methods i.e., they do not require "line-of-sight", are robust to occlusions and changes in illumination, and can function as a backup in case vision/lidar sensors fail. Therefore, audio-based methods can be useful even for applications in which only visual information is of interest Our framework is based on Manifold Learning and consists of two steps. First, we train a Vector-Quantized Variational Auto-Encoder to learn the data manifold of the particular visual modality we are interested in. Second, we train an Audio Transformation network to map multi-channel audio signals to the latent representation of the corresponding visual sample. We show that our method is able to produce meaningful images from audio using a publicly available audio/visual dataset. In particular, we consider the prediction of the following visual modalities from audio: depth and semantic segmentation. We hope the findings of our work can facilitate further research in visual information extraction from audio. Code is available at: https://github.com/ubc-vision/audio_manifold.
Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards
Authors: Hanwen Kang, Xing Wang
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Semantic segmentation is a fundamental task for agricultural robots to understand the surrounding environments in natural orchards. The recent development of the LiDAR techniques enables the robot to acquire accurate range measurements of the view in the unstructured orchards. Compared to RGB images, 3D point clouds have geometrical properties. By combining the LiDAR and camera, rich information on geometries and textures can be obtained. In this work, we propose a deep-learning-based segmentation method to perform accurate semantic segmentation on fused data from a LiDAR-Camera visual sensor. Two critical problems are explored and solved in this work. The first one is how to efficiently fused the texture and geometrical features from multi-sensor data. The second one is how to efficiently train the 3D segmentation network under severely imbalance class conditions. Moreover, an implementation of 3D segmentation in orchards including LiDAR-Camera data fusion, data collection and labelling, network training, and model inference is introduced in detail. In the experiment, we comprehensively analyze the network setup when dealing with highly unstructured and noisy point clouds acquired from an apple orchard. Overall, our proposed method achieves 86.2% mIoU on the segmentation of fruits on the high-resolution point cloud (100k-200k points). The experiment results show that the proposed method can perform accurate segmentation in real orchard environments.
A Cooperative Perception Environment for Traffic Operations and Control
Authors: Hanlin Chen, Brian Liu, Xumiao Zhang, Feng Qian, Z. Morley Mao, Yiheng Feng
Subjects: Robotics (cs.RO); Systems and Control (eess.SY)
Abstract
Existing data collection methods for traffic operations and control usually rely on infrastructure-based loop detectors or probe vehicle trajectories. Connected and automated vehicles (CAVs) not only can report data about themselves but also can provide the status of all detected surrounding vehicles. Integration of perception data from multiple CAVs as well as infrastructure sensors (e.g., LiDAR) can provide richer information even under a very low penetration rate. This paper aims to develop a cooperative data collection system, which integrates Lidar point cloud data from both infrastructure and CAVs to create a cooperative perception environment for various transportation applications. The state-of-the-art 3D detection models are applied to detect vehicles in the merged point cloud. We test the proposed cooperative perception environment with the max pressure adaptive signal control model in a co-simulation platform with CARLA and SUMO. Results show that very low penetration rates of CAV plus an infrastructure sensor are sufficient to achieve comparable performance with 30% or higher penetration rates of connected vehicles (CV). We also show the equivalent CV penetration rate (E-CVPR) under different CAV penetration rates to demonstrate the data collection efficiency of the cooperative perception environment.
Keyword: loop detection
There is no result
Keyword: nerf
360Roam: Real-Time Indoor Roaming Using Geometry-Aware ${360^\circ}$ Radiance Fields
Abstract
Neural radiance field (NeRF) has recently achieved impressive results in novel view synthesis. However, previous works on NeRF mainly focus on object-centric scenarios. In this work, we propose 360Roam, a novel scene-level NeRF system that can synthesize images of large-scale indoor scenes in real time and support VR roaming. Our system first builds an omnidirectional neural radiance field 360NeRF from multiple input $360^\circ$ images. Using 360NeRF, we then progressively estimate a 3D probabilistic occupancy map which represents the scene geometry in the form of spacial density. Skipping empty spaces and upsampling occupied voxels essentially allows us to accelerate volume rendering by using 360NeRF in a geometry-aware fashion. Furthermore, we use an adaptive divide-and-conquer strategy to slim and fine-tune the radiance fields for further improvement. The floorplan of the scene extracted from the occupancy map can provide guidance for ray sampling and facilitate a realistic roaming experience. To show the efficacy of our system, we collect a $360^\circ$ image dataset in a large variety of scenes and conduct extensive experiments. Quantitative and qualitative comparisons among baselines illustrated our predominant performance in novel view synthesis for complex indoor scenes.
Keyword: mapping
Learning Fast and Precise Pixel-to-Torque Control
Authors: Steffen Bleher, Steve Heim, Sebastian Trimpe
Subjects: Robotics (cs.RO); Systems and Control (eess.SY)
Abstract
In the field, robots often need to operate in unknown and unstructured environments, where accurate sensing and state estimation (SE) becomes a major challenge. Cameras have been used to great success in mapping and planning in such environments, as well as complex but quasi-static tasks such as grasping, but are rarely integrated into the control loop for unstable systems. Learning pixel-to-torque control promises to allow robots to flexibly handle a wider variety of tasks. Although they do not present additional theoretical obstacles, learning pixel-to-torque control for unstable systems that that require precise and high bandwidth control still poses a significant practical challenge, and best practices have not yet been established. To help drive reproducible research on the practical aspects of learning pixel-to-torque control, we propose a platform that can flexibly represent the entire process, from lab to deployment, for learning pixel-to-torque control on a robot with fast, unstable dynamics: the vision-based Furuta pendulum. The platform can be reproduced with either off-the-shelf or custom-built hardware. We expect that this platform will allow researchers to quickly and systematically test different approaches, as well as reproduce and benchmark case studies from other labs. We also present a first case study on this system using DNNs which, to the best of our knowledge, is the first demonstration of learning pixel-to-torque control on an unstable system with update rates faster than 100 Hz. A video synopsis can be found online at https://youtu.be/S2llScfG-8E, and in the supplementary material.
End-to-end deep learning for directly estimating grape yield from ground-based imagery
Authors: Alexander G. Olenskyj, Brent S. Sams, Zhenghao Fei, Vishal Singh, Pranav V. Raja, Gail M. Bornhorst, J. Mason Earles
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Yield estimation is a powerful tool in vineyard management, as it allows growers to fine-tune practices to optimize yield and quality. However, yield estimation is currently performed using manual sampling, which is time-consuming and imprecise. This study demonstrates the application of proximal imaging combined with deep learning for yield estimation in vineyards. Continuous data collection using a vehicle-mounted sensing kit combined with collection of ground truth yield data at harvest using a commercial yield monitor allowed for the generation of a large dataset of 23,581 yield points and 107,933 images. Moreover, this study was conducted in a mechanically managed commercial vineyard, representing a challenging environment for image analysis but a common set of conditions in the California Central Valley. Three model architectures were tested: object detection, CNN regression, and transformer models. The object detection model was trained on hand-labeled images to localize grape bunches, and either bunch count or pixel area was summed to correlate with grape yield. Conversely, regression models were trained end-to-end to predict grape yield from image data without the need for hand labeling. Results demonstrated that both a transformer as well as the object detection model with pixel area processing performed comparably, with a mean absolute percent error of 18% and 18.5%, respectively on a representative holdout dataset. Saliency mapping was used to demonstrate the attention of the CNN model was localized near the predicted location of grape bunches, as well as on the top of the grapevine canopy. Overall, the study showed the applicability of proximal imaging and deep learning for prediction of grapevine yield on a large scale. Additionally, the end-to-end modeling approach was able to perform comparably to the object detection approach while eliminating the need for hand-labeling.
Constructing Balance from Imbalance for Long-tailed Image Recognition
Authors: Yue Xu, Yong-Lu Li, Jiefeng Li, Cewu Lu
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Long-tailed image recognition presents massive challenges to deep learning systems since the imbalance between majority (head) classes and minority (tail) classes severely skews the data-driven deep neural networks. Previous methods tackle with data imbalance from the viewpoints of data distribution, feature space, and model design, etc.In this work, instead of directly learning a recognition model, we suggest confronting the bottleneck of head-to-tail bias before classifier learning, from the previously omitted perspective of balancing label space. To alleviate the head-to-tail bias, we propose a concise paradigm by progressively adjusting label space and dividing the head classes and tail classes, dynamically constructing balance from imbalance to facilitate the classification. With flexible data filtering and label space mapping, we can easily embed our approach to most classification models, especially the decoupled training methods. Besides, we find the separability of head-tail classes varies among different features with different inductive biases. Hence, our proposed model also provides a feature evaluation method and paves the way for long-tailed feature learning. Extensive experiments show that our method can boost the performance of state-of-the-arts of different types on widely-used benchmarks. Code is available at https://github.com/silicx/DLSA.
ZeroMesh: Zero-shot Single-view 3D Mesh Reconstruction
Authors: Xianghui Yang, Guosheng Lin, Luping Zhou
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Single-view 3D object reconstruction is a fundamental and challenging computer vision task that aims at recovering 3D shapes from single-view RGB images. Most existing deep learning based reconstruction methods are trained and evaluated on the same categories, and they cannot work well when handling objects from novel categories that are not seen during training. Focusing on this issue, this paper tackles Zero-shot Single-view 3D Mesh Reconstruction, to study the model generalization on unseen categories and encourage models to reconstruct objects literally. Specifically, we propose an end-to-end two-stage network, ZeroMesh, to break the category boundaries in reconstruction. Firstly, we factorize the complicated image-to-mesh mapping into two simpler mappings, i.e., image-to-point mapping and point-to-mesh mapping, while the latter is mainly a geometric problem and less dependent on object categories. Secondly, we devise a local feature sampling strategy in 2D and 3D feature spaces to capture the local geometry shared across objects to enhance model generalization. Thirdly, apart from the traditional point-to-point supervision, we introduce a multi-view silhouette loss to supervise the surface generation process, which provides additional regularization and further relieves the overfitting problem. The experimental results show that our method significantly outperforms the existing works on the ShapeNet and Pix3D under different scenarios and various metrics, especially for novel objects.
Relict landslide detection in rainforest areas using a combination of k-means clustering algorithm and Deep-Learning semantic segmentation models
Authors: Guilherme P.B. Garcia, Carlos H. Grohmann, Lucas P. Soares, Mateus Espadoto
Subjects: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV); Data Analysis, Statistics and Probability (physics.data-an)
Abstract
Landslides are destructive and recurrent natural disasters on steep slopes and represent a risk to lives and properties. Knowledge of relict landslides' location is vital to understand their mechanisms, update inventory maps and improve risk assessment. However, relict landslide mapping is complex in tropical regions covered with rainforest vegetation. A new CNN approach is proposed for semi-automatic detection of relict landslides, which uses a dataset generated by a k-means clustering algorithm and has a pre-training step. The weights computed in the pre-training are used to fine-tune the CNN training process. A comparison between the proposed and standard approaches is performed using CBERS-4A WPM images. Three CNNs for semantic segmentation are used (U-Net, FPN, Linknet) with two augmented datasets. A total of 42 combinations of CNNs are tested. Values of precision and recall were very similar between the combinations tested. Recall was higher than 75\% for every combination, but precision values were usually smaller than 20\%. False positives (FP) samples were addressed as the cause for these low precision values. Predictions of the proposed approach were more accurate and correctly detected more landslides. This work demonstrates that there are limitations for detecting relict landslides in areas covered with rainforest, mainly related to similarities between the spectral response of pastures and deforested areas with \textit{Gleichenella sp.} ferns, commonly used as an indicator of landslide scars.
QC-ODKLA: Quantized and Communication-Censored Online Decentralized Kernel Learning via Linearized ADMM
Abstract
This paper focuses on online kernel learning over a decentralized network. Each agent in the network receives continuous streaming data locally and works collaboratively to learn a nonlinear prediction function that is globally optimal in the reproducing kernel Hilbert space with respect to the total instantaneous costs of all agents. In order to circumvent the curse of dimensionality issue in traditional online kernel learning, we utilize random feature (RF) mapping to convert the non-parametric kernel learning problem into a fixed-length parametric one in the RF space. We then propose a novel learning framework named Online Decentralized Kernel learning via Linearized ADMM (ODKLA) to efficiently solve the online decentralized kernel learning problem. To further improve the communication efficiency, we add the quantization and censoring strategies in the communication stage and develop the Quantized and Communication-censored ODKLA (QC-ODKLA) algorithm. We theoretically prove that both ODKLA and QC-ODKLA can achieve the optimal sublinear regret $\mathcal{O}(\sqrt{T})$ over $T$ time slots. Through numerical experiments, we evaluate the learning effectiveness, communication, and computation efficiencies of the proposed methods.
Transformers as Meta-Learners for Implicit Neural Representations
Authors: Yinbo Chen, Xiaolong Wang
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Abstract
Implicit Neural Representations (INRs) have emerged and shown their benefits over discrete representations in recent years. However, fitting an INR to the given observations usually requires optimization with gradient descent from scratch, which is inefficient and does not generalize well with sparse observations. To address this problem, most of the prior works train a hypernetwork that generates a single vector to modulate the INR weights, where the single vector becomes an information bottleneck that limits the reconstruction precision of the output INR. Recent work shows that the whole set of weights in INR can be precisely inferred without the single-vector bottleneck by gradient-based meta-learning. Motivated by a generalized formulation of gradient-based meta-learning, we propose a formulation that uses Transformers as hypernetworks for INRs, where it can directly build the whole set of INR weights with Transformers specialized as set-to-set mapping. We demonstrate the effectiveness of our method for building INRs in different tasks and domains, including 2D image regression and view synthesis for 3D objects. Our work draws connections between the Transformer hypernetworks and gradient-based meta-learning algorithms and we provide further analysis for understanding the generated INRs. The project page with code is at \url{https://yinboc.github.io/trans-inr/} .
Keyword: localization
AG2U -- Autonomous Grading Under Uncertainties
Authors: Yakov Miron, Yuval Goldfracht, Chana Ross, Dotan Di Castro, Itzik Klein
Abstract
Surface grading, the process of leveling an uneven area containing pre-dumped sand piles, is an important task in the construction site pipeline. This labour-intensive process is often carried out by a dozer, a key machinery tool at any construction site. Current attempts to automate surface grading assume perfect localization. However, in real-world scenarios, this assumption fails, as agents are presented with imperfect perception, which leads to degraded performance. In this work, we address the problem of autonomous grading under uncertainties. First, we implement a simulation and a scaled real-world prototype environment to enable rapid policy exploration and evaluation in this setting. Second, we formalize the problem as a partially observable markov decision process and train an agent capable of handling such uncertainties. We show, through rigorous experiments, that an agent trained under perfect localization will suffer degraded performance when presented with localization uncertainties. However, an agent trained using our method will develop a more robust policy for addressing such errors and, consequently, exhibit a better grading performance.
A Sequential MUSIC algorithm for Scatterers Detection 2 in SAR Tomography Enhanced by a Robust Covariance 3 Estimator
Authors: Ahmad Naghavi, Mohammad Sadegh Fazel, Mojtaba Beheshti, Ehsan Yazdian
Subjects: Information Theory (cs.IT); Signal Processing (eess.SP)
Abstract
Synthetic aperture radar (SAR) tomography (TomoSAR) is an appealing tool for the extraction of height information of urban infrastructures. Due to the widespread applications of the MUSIC algorithm in source localization, it is a suitable solution in TomoSAR when multiple snapshots (looks) are available. While the classical MUSIC algorithm aims to estimate the whole reflectivity profile of scatterers, sequential MUSIC algorithms are suited for the detection of sparse point-like scatterers. In this class of methods, successive cancellation is performed through orthogonal complement projections on the MUSIC power spectrum. In this work, a new sequential MUSIC algorithm named recursive covariance canceled MUSIC (RCC-MUSIC), is proposed. This method brings higher accuracy in comparison with the previous sequential methods at the cost of a negligible increase in computational cost. Furthermore, to improve the performance of RCC-MUSIC, it is combined with the recent method of covariance matrix estimation called correlation subspace. Utilizing the correlation subspace method results in a denoised covariance matrix which in turn, increases the accuracy of subspace-based methods. Several numerical examples are presented to compare the performance of the proposed method with the relevant state-of-the-art methods. As a subspace method, simulation results demonstrate the efficiency of the proposed method in terms of estimation accuracy and computational load.
Keyword: transformer
HiCu: Leveraging Hierarchy for Curriculum Learning in Automated ICD Coding
Abstract
There are several opportunities for automation in healthcare that can improve clinician throughput. One such example is assistive tools to document diagnosis codes when clinicians write notes. We study the automation of medical code prediction using curriculum learning, which is a training strategy for machine learning models that gradually increases the hardness of the learning tasks from easy to difficult. One of the challenges in curriculum learning is the design of curricula -- i.e., in the sequential design of tasks that gradually increase in difficulty. We propose Hierarchical Curriculum Learning (HiCu), an algorithm that uses graph structure in the space of outputs to design curricula for multi-label classification. We create curricula for multi-label classification models that predict ICD diagnosis and procedure codes from natural language descriptions of patients. By leveraging the hierarchy of ICD codes, which groups diagnosis codes based on various organ systems in the human body, we find that our proposed curricula improve the generalization of neural network-based predictive models across recurrent, convolutional, and transformer-based architectures. Our code is available at https://github.com/wren93/HiCu-ICD.
End-to-end deep learning for directly estimating grape yield from ground-based imagery
Authors: Alexander G. Olenskyj, Brent S. Sams, Zhenghao Fei, Vishal Singh, Pranav V. Raja, Gail M. Bornhorst, J. Mason Earles
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Yield estimation is a powerful tool in vineyard management, as it allows growers to fine-tune practices to optimize yield and quality. However, yield estimation is currently performed using manual sampling, which is time-consuming and imprecise. This study demonstrates the application of proximal imaging combined with deep learning for yield estimation in vineyards. Continuous data collection using a vehicle-mounted sensing kit combined with collection of ground truth yield data at harvest using a commercial yield monitor allowed for the generation of a large dataset of 23,581 yield points and 107,933 images. Moreover, this study was conducted in a mechanically managed commercial vineyard, representing a challenging environment for image analysis but a common set of conditions in the California Central Valley. Three model architectures were tested: object detection, CNN regression, and transformer models. The object detection model was trained on hand-labeled images to localize grape bunches, and either bunch count or pixel area was summed to correlate with grape yield. Conversely, regression models were trained end-to-end to predict grape yield from image data without the need for hand labeling. Results demonstrated that both a transformer as well as the object detection model with pixel area processing performed comparably, with a mean absolute percent error of 18% and 18.5%, respectively on a representative holdout dataset. Saliency mapping was used to demonstrate the attention of the CNN model was localized near the predicted location of grape bunches, as well as on the top of the grapevine canopy. Overall, the study showed the applicability of proximal imaging and deep learning for prediction of grapevine yield on a large scale. Additionally, the end-to-end modeling approach was able to perform comparably to the object detection approach while eliminating the need for hand-labeling.
MVSFormer: Learning Robust Image Representations via Transformers and Temperature-based Depth for Multi-View Stereo
Authors: Chenjie Cao, Xinlin Ren, Yanwei Fu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Feature representation learning is the key recipe for learning-based Multi-View Stereo (MVS). As the common feature extractor of learning-based MVS, vanilla Feature Pyramid Networks (FPN) suffers from discouraged feature representations for reflection and texture-less areas, which limits the generalization of MVS. Even FPNs worked with pre-trained Convolutional Neural Networks (CNNs) fail to tackle these issues. On the other hand, Vision Transformers (ViTs) have achieved prominent success in many 2D vision tasks. Thus we ask whether ViTs can facilitate the feature learning in MVS? In this paper, we propose a pre-trained ViT enhanced MVS network called MVSFormer, which can learn more reliable feature representations benefited by informative priors from ViT. Then MVSFormer-P and MVSFormer-H are further proposed with fixed ViT weights and trainable ones respectively. MVSFormer-P is more efficient while MVSFormer-H can achieve superior performance. To make ViTs robust to arbitrary resolutions for MVS tasks, we propose to use an efficient multi-scale training with gradient accumulation. Moreover, we discuss the merits and drawbacks of classification and regression-based MVS methods, and further propose to unify them with a temperature-based strategy. MVSFormer achieves state-of-the-art performance on the DTU dataset. Particularly, our anonymous submission of MVSFormer is ranked in the Top-1 position on both intermediate and advanced sets of the highly competitive Tanks-and-Temples leaderboard on the day of submission compared with other published works. Codes and models will be released.
ACSGRegNet: A Deep Learning-based Framework for Unsupervised Joint Affine and Diffeomorphic Registration of Lumbar Spine CT via Cross- and Self-Attention Fusion
Authors: Xiaoru Gao, GuoYan Zheng
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Registration plays an important role in medical image analysis. Deep learning-based methods have been studied for medical image registration, which leverage convolutional neural networks (CNNs) for efficiently regressing a dense deformation field from a pair of images. However, CNNs are limited in its ability to extract semantically meaningful intra- and inter-image spatial correspondences, which are of importance for accurate image registration. This study proposes a novel end-to-end deep learning-based framework for unsupervised affine and diffeomorphic deformable registration, referred as ACSGRegNet, which integrates a cross-attention module for establishing inter-image feature correspondences and a self-attention module for intra-image anatomical structures aware. Both attention modules are built on transformer encoders. The output from each attention module is respectively fed to a decoder to generate a velocity field. We further introduce a gated fusion module to fuse both velocity fields. The fused velocity field is then integrated to a dense deformation field. Extensive experiments are conducted on lumbar spine CT images. Once the model is trained, pairs of unseen lumbar vertebrae can be registered in one shot. Evaluated on 450 pairs of vertebral CT data, our method achieved an average Dice of 0.963 and an average distance error of 0.321mm, which are better than the state-of-the-art (SOTA).
DropKey
Authors: Bonan Li, Yinhan Hu, Xuecheng Nie, Congying Han, Xiangjian Jiang, Tiande Guo, Luoqi Liu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
In this paper, we focus on analyzing and improving the dropout technique for self-attention layers of Vision Transformer, which is important while surprisingly ignored by prior works. In particular, we conduct researches on three core questions: First, what to drop in self-attention layers? Different from dropping attention weights in literature, we propose to move dropout operations forward ahead of attention matrix calculation and set the Key as the dropout unit, yielding a novel dropout-before-softmax scheme. We theoretically verify that this scheme helps keep both regularization and probability features of attention weights, alleviating the overfittings problem to specific patterns and enhancing the model to globally capture vital information; Second, how to schedule the drop ratio in consecutive layers? In contrast to exploit a constant drop ratio for all layers, we present a new decreasing schedule that gradually decreases the drop ratio along the stack of self-attention layers. We experimentally validate the proposed schedule can avoid overfittings in low-level features and missing in high-level semantics, thus improving the robustness and stableness of model training; Third, whether need to perform structured dropout operation as CNN? We attempt patch-based block-version of dropout operation and find that this useful trick for CNN is not essential for ViT. Given exploration on the above three questions, we present the novel DropKey method that regards Key as the drop unit and exploits decreasing schedule for drop ratio, improving ViTs in a general way. Comprehensive experiments demonstrate the effectiveness of DropKey for various ViT architectures, \emph{e.g.} T2T and VOLO, as well as for various vision tasks, \emph{e.g.}, image classification, object detection, human-object interaction detection and human body shape recovery. Codes will be released upon acceptance.
UTOPIC: Uncertainty-aware Overlap Prediction Network for Partial Point Cloud Registration
Abstract
High-confidence overlap prediction and accurate correspondences are critical for cutting-edge models to align paired point clouds in a partial-to-partial manner. However, there inherently exists uncertainty between the overlapping and non-overlapping regions, which has always been neglected and significantly affects the registration performance. Beyond the current wisdom, we propose a novel uncertainty-aware overlap prediction network, dubbed UTOPIC, to tackle the ambiguous overlap prediction problem; to our knowledge, this is the first to explicitly introduce overlap uncertainty to point cloud registration. Moreover, we induce the feature extractor to implicitly perceive the shape knowledge through a completion decoder, and present a geometric relation embedding for Transformer to obtain transformation-invariant geometry-aware feature representations. With the merits of more reliable overlap scores and more precise dense correspondences, UTOPIC can achieve stable and accurate registration results, even for the inputs with limited overlapping areas. Extensive quantitative and qualitative experiments on synthetic and real benchmarks demonstrate the superiority of our approach over state-of-the-art methods. Code is available at https://github.com/ZhileiChen99/UTOPIC.
Integrating Knowledge Graph embedding and pretrained Language Models in Hypercomplex Spaces
Authors: Mojtaba Nayyeri, Zihao Wang, Mst. Mahfuja Akter, Mirza Mohtashim Alam, Md Rashad Al Hasan Rony, Jens Lehmann, Steffen Staab
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Knowledge Graphs, such as Wikidata, comprise structural and textual knowledge in order to represent knowledge. For each of the two modalities dedicated approaches for graph embedding and language models learn patterns that allow for predicting novel structural knowledge. Few approaches have integrated learning and inference with both modalities and these existing ones could only partially exploit the interaction of structural and textual knowledge. In our approach, we build on existing strong representations of single modalities and we use hypercomplex algebra to represent both, (i), single-modality embedding as well as, (ii), the interaction between different modalities and their complementary means of knowledge representation. More specifically, we suggest Dihedron and Quaternion representations of 4D hypercomplex numbers to integrate four modalities namely structural knowledge graph embedding, word-level representations (e.g.\ Word2vec, Fasttext), sentence-level representations (Sentence transformer), and document-level representations (sentence transformer, Doc2vec). Our unified vector representation scores the plausibility of labelled edges via Hamilton and Dihedron products, thus modeling pairwise interactions between different modalities. Extensive experimental evaluation on standard benchmark datasets shows the superiority of our two new models using abundant textual information besides sparse structural knowledge to enhance performance in link prediction tasks.
Transformers as Meta-Learners for Implicit Neural Representations
Authors: Yinbo Chen, Xiaolong Wang
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Abstract
Implicit Neural Representations (INRs) have emerged and shown their benefits over discrete representations in recent years. However, fitting an INR to the given observations usually requires optimization with gradient descent from scratch, which is inefficient and does not generalize well with sparse observations. To address this problem, most of the prior works train a hypernetwork that generates a single vector to modulate the INR weights, where the single vector becomes an information bottleneck that limits the reconstruction precision of the output INR. Recent work shows that the whole set of weights in INR can be precisely inferred without the single-vector bottleneck by gradient-based meta-learning. Motivated by a generalized formulation of gradient-based meta-learning, we propose a formulation that uses Transformers as hypernetworks for INRs, where it can directly build the whole set of INR weights with Transformers specialized as set-to-set mapping. We demonstrate the effectiveness of our method for building INRs in different tasks and domains, including 2D image regression and view synthesis for 3D objects. Our work draws connections between the Transformer hypernetworks and gradient-based meta-learning algorithms and we provide further analysis for understanding the generated INRs. The project page with code is at \url{https://yinboc.github.io/trans-inr/} .
Keyword: autonomous driving
Reinforcement Learning for Joint V2I Network Selection and Autonomous Driving Policies
Abstract
Vehicle-to-Infrastructure (V2I) communication is becoming critical for the enhanced reliability of autonomous vehicles (AVs). However, the uncertainties in the road-traffic and AVs' wireless connections can severely impair timely decision-making. It is thus critical to simultaneously optimize the AVs' network selection and driving policies in order to minimize road collisions while maximizing the communication data rates. In this paper, we develop a reinforcement learning (RL) framework to characterize efficient network selection and autonomous driving policies in a multi-band vehicular network (VNet) operating on conventional sub-6GHz spectrum and Terahertz (THz) frequencies. The proposed framework is designed to (i) maximize the traffic flow and minimize collisions by controlling the vehicle's motion dynamics (i.e., speed and acceleration) from autonomous driving perspective, and (ii) maximize the data rates and minimize handoffs by jointly controlling the vehicle's motion dynamics and network selection from telecommunication perspective. We cast this problem as a Markov Decision Process (MDP) and develop a deep Q-learning based solution to optimize the actions such as acceleration, deceleration, lane-changes, and AV-base station assignments for a given AV's state. The AV's state is defined based on the velocities and communication channel states of AVs. Numerical results demonstrate interesting insights related to the inter-dependency of vehicle's motion dynamics, handoffs, and the communication data rate. The proposed policies enable AVs to adopt safe driving behaviors with improved connectivity.
LaneSNNs: Spiking Neural Networks for Lane Detection on the Loihi Neuromorphic Processor
Authors: Alberto Viale, Alberto Marchisio, Maurizio Martina, Guido Masera, Muhammad Shafique
Subjects: Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG)
Abstract
Autonomous Driving (AD) related features represent important elements for the next generation of mobile robots and autonomous vehicles focused on increasingly intelligent, autonomous, and interconnected systems. The applications involving the use of these features must provide, by definition, real-time decisions, and this property is key to avoid catastrophic accidents. Moreover, all the decision processes must require low power consumption, to increase the lifetime and autonomy of battery-driven systems. These challenges can be addressed through efficient implementations of Spiking Neural Networks (SNNs) on Neuromorphic Chips and the use of event-based cameras instead of traditional frame-based cameras. In this paper, we present a new SNN-based approach, called LaneSNN, for detecting the lanes marked on the streets using the event-based camera input. We develop four novel SNN models characterized by low complexity and fast response, and train them using an offline supervised learning rule. Afterward, we implement and map the learned SNNs models onto the Intel Loihi Neuromorphic Research Chip. For the loss function, we develop a novel method based on the linear composition of Weighted binary Cross Entropy (WCE) and Mean Squared Error (MSE) measures. Our experimental results show a maximum Intersection over Union (IoU) measure of about 0.62 and very low power consumption of about 1 W. The best IoU is achieved with an SNN implementation that occupies only 36 neurocores on the Loihi processor while providing a low latency of less than 8 ms to recognize an image, thereby enabling real-time performance. The IoU measures provided by our networks are comparable with the state-of-the-art, but at a much low power consumption of 1 W.
IPDAE: Improved Patch-Based Deep Autoencoder for Lossy Point Cloud Geometry Compression
Authors: Kang You, Pan Gao, Qing Li
Subjects: Computer Vision and Pattern Recognition (cs.CV); Information Theory (cs.IT); Multimedia (cs.MM); Image and Video Processing (eess.IV)
Abstract
Point cloud is a crucial representation of 3D contents, which has been widely used in many areas such as virtual reality, mixed reality, autonomous driving, etc. With the boost of the number of points in the data, how to efficiently compress point cloud becomes a challenging problem. In this paper, we propose a set of significant improvements to patch-based point cloud compression, i.e., a learnable context model for entropy coding, octree coding for sampling centroid points, and an integrated compression and training process. In addition, we propose an adversarial network to improve the uniformity of points during reconstruction. Our experiments show that the improved patch-based autoencoder outperforms the state-of-the-art in terms of rate-distortion performance, on both sparse and large-scale point clouds. More importantly, our method can maintain a short compression time while ensuring the reconstruction quality.
New submissions for Fri, 5 Aug 22
Keyword: SLAM
SROS2: Usable Cyber Security Tools for ROS 2
Keyword: odometry
There is no result
Keyword: livox
There is no result
Keyword: loam
There is no result
Keyword: lidar
Estimating Visual Information From Audio Through Manifold Learning
Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards
A Cooperative Perception Environment for Traffic Operations and Control
Keyword: loop detection
There is no result
Keyword: nerf
360Roam: Real-Time Indoor Roaming Using Geometry-Aware ${360^\circ}$ Radiance Fields
Keyword: mapping
Learning Fast and Precise Pixel-to-Torque Control
End-to-end deep learning for directly estimating grape yield from ground-based imagery
Constructing Balance from Imbalance for Long-tailed Image Recognition
ZeroMesh: Zero-shot Single-view 3D Mesh Reconstruction
Relict landslide detection in rainforest areas using a combination of k-means clustering algorithm and Deep-Learning semantic segmentation models
QC-ODKLA: Quantized and Communication-Censored Online Decentralized Kernel Learning via Linearized ADMM
Transformers as Meta-Learners for Implicit Neural Representations
Keyword: localization
AG2U -- Autonomous Grading Under Uncertainties
A Sequential MUSIC algorithm for Scatterers Detection 2 in SAR Tomography Enhanced by a Robust Covariance 3 Estimator
Keyword: transformer
HiCu: Leveraging Hierarchy for Curriculum Learning in Automated ICD Coding
End-to-end deep learning for directly estimating grape yield from ground-based imagery
MVSFormer: Learning Robust Image Representations via Transformers and Temperature-based Depth for Multi-View Stereo
ACSGRegNet: A Deep Learning-based Framework for Unsupervised Joint Affine and Diffeomorphic Registration of Lumbar Spine CT via Cross- and Self-Attention Fusion
DropKey
UTOPIC: Uncertainty-aware Overlap Prediction Network for Partial Point Cloud Registration
Integrating Knowledge Graph embedding and pretrained Language Models in Hypercomplex Spaces
Transformers as Meta-Learners for Implicit Neural Representations
Keyword: autonomous driving
Reinforcement Learning for Joint V2I Network Selection and Autonomous Driving Policies
LaneSNNs: Spiking Neural Networks for Lane Detection on the Loihi Neuromorphic Processor
IPDAE: Improved Patch-Based Deep Autoencoder for Lossy Point Cloud Geometry Compression