Accurate 3D object detection (3DOD) is crucial for safe navigation of complex environments by autonomous robots. Regressing accurate 3D bounding boxes in cluttered environments based on sparse LiDAR data is however a highly challenging problem. We address this task by exploring recent advances in conditional energy-based models (EBMs) for probabilistic regression. While methods employing EBMs for regression have demonstrated impressive performance on 2D object detection in images, these techniques are not directly applicable to 3D bounding boxes. In this work, we therefore design a differentiable pooling operator for 3D bounding boxes, serving as the core module of our EBM network. We further integrate this general approach into the state-of-the-art 3D object detector SA-SSD. On the KITTI dataset, our proposed approach consistently outperforms the SA-SSD baseline across all 3DOD metrics, demonstrating the potential of EBM-based regression for highly accurate 3DOD. Code is available at //github.com/fregu856/ebms_3dod.
相關內容
Unsupervised depth completion methods are trained by minimizing sparse depth and image reconstruction error. Block artifacts from resampling, intensity saturation, and occlusions are amongst the many undesirable by-products of common data augmentation schemes that affect image reconstruction quality, and thus the training signal. Hence, typical augmentations on images viewed as essential to training pipelines in other vision tasks have seen limited use beyond small image intensity changes and flipping. The sparse depth modality have seen even less as intensity transformations alter the scale of the 3D scene, and geometric transformations may decimate the sparse points during resampling. We propose a method that unlocks a wide range of previously-infeasible geometric augmentations for unsupervised depth completion. This is achieved by reversing, or ``undo"-ing, geometric transformations to the coordinates of the output depth, warping the depth map back to the original reference frame. This enables computing the reconstruction losses using the original images and sparse depth maps, eliminating the pitfalls of naive loss computation on the augmented inputs. This simple yet effective strategy allows us to scale up augmentations to boost performance. We demonstrate our method on indoor (VOID) and outdoor (KITTI) datasets where we improve upon three existing methods by an average of 11.75% across both datasets.
The mobile robot relies on SLAM (Simultaneous Localization and Mapping) to provide autonomous navigation and task execution in complex and unknown environments. However, it is hard to develop a dedicated algorithm for mobile robots due to dynamic and challenging situations, such as poor lighting conditions and motion blur. To tackle this issue, we propose a tightly-coupled LiDAR-visual SLAM based on geometric features, which includes two sub-systems (LiDAR and monocular visual SLAM) and a fusion framework. The fusion framework associates the depth and semantics of the multi-modal geometric features to complement the visual line landmarks and to add direction optimization in Bundle Adjustment (BA). This further constrains visual odometry. On the other hand, the entire line segment detected by the visual subsystem overcomes the limitation of the LiDAR subsystem, which can only perform the local calculation for geometric features. It adjusts the direction of linear feature points and filters out outliers, leading to a higher accurate odometry system. Finally, we employ a module to detect the subsystem's operation, providing the LiDAR subsystem's output as a complementary trajectory to our system while visual subsystem tracking fails. The evaluation results on the public dataset M2DGR, gathered from ground robots across various indoor and outdoor scenarios, show that our system achieves more accurate and robust pose estimation compared to current state-of-the-art multi-modal methods.
In advanced human-robot interaction tasks, visual target navigation is crucial for autonomous robots navigating unknown environments. While numerous approaches have been developed in the past, most are designed for single-robot operations, which often suffer from reduced efficiency and robustness due to environmental complexities. Furthermore, learning policies for multi-robot collaboration are resource-intensive. To address these challenges, we propose Co-NavGPT, an innovative framework that integrates Large Language Models (LLMs) as a global planner for multi-robot cooperative visual target navigation. Co-NavGPT encodes the explored environment data into prompts, enhancing LLMs' scene comprehension. It then assigns exploration frontiers to each robot for efficient target search. Experimental results on Habitat-Matterport 3D (HM3D) demonstrate that Co-NavGPT surpasses existing models in success rates and efficiency without any learning process, demonstrating the vast potential of LLMs in multi-robot collaboration domains. The supplementary video, prompts, and code can be accessed via the following link: //sites.google.com/view/co-navgpt
User attribute prediction is a crucial task in various industries. However, sharing user data across different organizations faces challenges due to privacy concerns and legal requirements regarding personally identifiable information. Regulations such as the General Data Protection Regulation (GDPR) in the European Union and the Personal Information Protection Law of the People's Republic of China impose restrictions on data sharing. To address the need for utilizing features from multiple clients while adhering to legal requirements, federated learning algorithms have been proposed. These algorithms aim to predict user attributes without directly sharing the data. However, existing approaches typically rely on matching users across companies, which can result in dishonest partners discovering user lists or the inability to utilize all available features. In this paper, we propose a novel algorithm for predicting user attributes without requiring user matching. Our approach involves training deep matrix factorization models on different clients and sharing only the item vectors. This allows us to predict user attributes without sharing the user vectors themselves. The algorithm is evaluated using the publicly available MovieLens dataset and demonstrate that it achieves similar performance to the FedAvg algorithm, reaching 96% of a single model's accuracy. The proposed algorithm is particularly well-suited for improving customer targeting and enhancing the overall customer experience. This paper presents a valuable contribution to the field of user attribute prediction by offering a novel algorithm that addresses some of the most pressing privacy concerns in this area.
When manipulating an object to accomplish complex tasks, humans rely on both vision and touch to keep track of the object's 6D pose. However, most existing object pose tracking systems in robotics rely exclusively on visual signals, which hinder a robot's ability to manipulate objects effectively. To address this limitation, we introduce TEG-Track, a tactile-enhanced 6D pose tracking system that can track previously unseen objects held in hand. From consecutive tactile signals, TEG-Track optimizes object velocities from marker flows when slippage does not occur, or regresses velocities using a slippage estimation network when slippage is detected. The estimated object velocities are integrated into a geometric-kinematic optimization scheme to enhance existing visual pose trackers. To evaluate our method and to facilitate future research, we construct a real-world dataset for visual-tactile in-hand object pose tracking. Experimental results demonstrate that TEG-Track consistently enhances state-of-the-art generalizable 6D pose trackers in synthetic and real-world scenarios. Our code and dataset are available at //github.com/leolyliu/TEG-Track.
Completion problems, of recovering points from a set of observed coordinates, are abundant in applications to image reconstruction, phylogenetics, and data science. We consider a completion problem coming from algebraic statistics, of determining those observed probabilities which can be finitely completed to a probability distribution in a given log--linear model. These observed probabilities either have a unique completion or two completions to the log--linear model depending on the set of observed coordinates.
This tutorial aims to establish connections between polynomial modular multiplication over a ring to circular convolution and discrete Fourier transform (DFT). The main goal is to extend the well-known theory of DFT in signal processing (SP) to other applications involving polynomials in a ring such as homomorphic encryption (HE). HE allows any third party to operate on the encrypted data without decrypting it in advance. Since most HE schemes are constructed from the ring-learning with errors (R-LWE) problem, efficient polynomial modular multiplication implementation becomes critical. Any improvement in the execution of these building blocks would have significant consequences for the global performance of HE. This lecture note describes three approaches to implementing long polynomial modular multiplication using the number theoretic transform (NTT): zero-padded convolution, without zero-padding, also referred to as negative wrapped convolution (NWC), and low-complexity NWC (LC-NWC).
Few-shot object detection (FSOD) benchmarks have advanced techniques for detecting new categories with limited annotations. Existing benchmarks repurpose well-established datasets like COCO by partitioning categories into base and novel classes for pre-training and fine-tuning respectively. However, these benchmarks do not reflect how FSOD is deployed in practice. Rather than only pre-training on a small number of base categories, we argue that it is more practical to fine-tune a foundation model (e.g., a vision-language model (VLM) pre-trained on web-scale data) for a target domain. Surprisingly, we find that zero-shot inference from VLMs like GroundingDINO significantly outperforms the state-of-the-art (48.3 vs. 33.1 AP) on COCO. However, such zero-shot models can still be misaligned to target concepts of interest. For example, trailers on the web may be different from trailers in the context of autonomous vehicles. In this work, we propose Foundational FSOD, a new benchmark protocol that evaluates detectors pre-trained on any external datasets and fine-tuned on K-shots per target class. Further, we note that current FSOD benchmarks are actually federated datasets containing exhaustive annotations for each category on a subset of the data. We leverage this insight to propose simple strategies for fine-tuning VLMs with federated losses. We demonstrate the effectiveness of our approach on LVIS and nuImages, improving over prior work by 5.9 AP.
Image-to-image translation aims to learn the mapping between two visual domains. There are two main challenges for many applications: 1) the lack of aligned training pairs and 2) multiple possible outputs from a single input image. In this work, we present an approach based on disentangled representation for producing diverse outputs without paired training images. To achieve diversity, we propose to embed images onto two spaces: a domain-invariant content space capturing shared information across domains and a domain-specific attribute space. Our model takes the encoded content features extracted from a given input and the attribute vectors sampled from the attribute space to produce diverse outputs at test time. To handle unpaired training data, we introduce a novel cross-cycle consistency loss based on disentangled representations. Qualitative results show that our model can generate diverse and realistic images on a wide range of tasks without paired training data. For quantitative comparisons, we measure realism with user study and diversity with a perceptual distance metric. We apply the proposed model to domain adaptation and show competitive performance when compared to the state-of-the-art on the MNIST-M and the LineMod datasets.
The low resolution of objects of interest in aerial images makes pedestrian detection and action detection extremely challenging tasks. Furthermore, using deep convolutional neural networks to process large images can be demanding in terms of computational requirements. In order to alleviate these challenges, we propose a two-step, yes and no question answering framework to find specific individuals doing one or multiple specific actions in aerial images. First, a deep object detector, Single Shot Multibox Detector (SSD), is used to generate object proposals from small aerial images. Second, another deep network, is used to learn a latent common sub-space which associates the high resolution aerial imagery and the pedestrian action labels that are provided by the human-based sources
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191