We present PVSeRF, a learning framework that reconstructs neural radiance fields from single-view RGB images, for novel view synthesis. Previous solutions, such as pixelNeRF, rely only on pixel-aligned features and suffer from feature ambiguity issues. As a result, they struggle with the disentanglement of geometry and appearance, leading to implausible geometries and blurry results. To address this challenge, we propose to incorporate explicit geometry reasoning and combine it with pixel-aligned features for radiance field prediction. Specifically, in addition to pixel-aligned features, we further constrain the radiance field learning to be conditioned on i) voxel-aligned features learned from a coarse volumetric grid and ii) fine surface-aligned features extracted from a regressed point cloud. We show that the introduction of such geometry-aware features helps to achieve a better disentanglement between appearance and geometry, i.e. recovering more accurate geometries and synthesizing higher quality images of novel views. Extensive experiments against state-of-the-art methods on ShapeNet benchmarks demonstrate the superiority of our approach for single-image novel view synthesis.
This letter studies a vertical federated edge learning (FEEL) system for collaborative objects/human motion recognition by exploiting the distributed integrated sensing and communication (ISAC). In this system, distributed edge devices first send wireless signals to sense targeted objects/human, and then exchange intermediate computed vectors (instead of raw sensing data) for collaborative recognition while preserving data privacy. To boost the spectrum and hardware utilization efficiency for FEEL, we exploit ISAC for both target sensing and data exchange, by employing dedicated frequency-modulated continuous-wave (FMCW) signals at each edge device. Under this setup, we propose a vertical FEEL framework for realizing the recognition based on the collected multi-view wireless sensing data. In this framework, each edge device owns an individual local L-model to transform its sensing data into an intermediate vector with relatively low dimensions, which is then transmitted to a coordinating edge device for final output via a common downstream S-model. By considering a human motion recognition task, experimental results show that our vertical FEEL based approach achieves recognition accuracy up to 98\% with an improvement up to 8\% compared to the benchmarks, including on-device training and horizontal FEEL.
3D scene understanding is a relatively emerging research field. In this paper, we introduce the Visual Question Answering task in 3D real-world scenes (VQA-3D), which aims to answer all possible questions given a 3D scene. To tackle this problem, the first VQA-3D dataset, namely CLEVR3D, is proposed, which contains 60K questions in 1,129 real-world scenes. Specifically, we develop a question engine leveraging 3D scene graph structures to generate diverse reasoning questions, covering the questions of objects' attributes (i.e., size, color, and material) and their spatial relationships. Built upon this dataset, we further design the first VQA-3D baseline model, TransVQA3D. The TransVQA3D model adopts well-designed Transformer architectures to achieve superior VQA-3D performance, compared with the pure language baseline and previous 3D reasoning methods directly applied to 3D scenarios. Experimental results verify that taking VQA-3D as an auxiliary task can boost the performance of 3D scene understanding, including scene graph analysis for the node-wise classification and whole-graph recognition.
Federated learning is a distributed machine learning mechanism where local devices collaboratively train a shared global model under the orchestration of a central server, while keeping all private data decentralized. In the system, model parameters and its updates are transmitted instead of raw data, and thus the communication bottleneck has become a key challenge. Besides, recent larger and deeper machine learning models also pose more difficulties in deploying them in a federated environment. In this paper, we design a federated two-stage learning framework that augments prototypical federated learning with a cut layer on devices and uses sign-based stochastic gradient descent with the majority vote method on model updates. Cut layer on devices learns informative and low-dimension representations of raw data locally, which helps reduce global model parameters and prevents data leakage. Sign-based SGD with the majority vote method for model updates also helps alleviate communication limitations. Empirically, we show that our system is an efficient and privacy preserving federated learning scheme and suits for general application scenarios.
This paper considers device-free sensing in an orthogonal frequency division multiplexing (OFDM) cellular network to enable integrated sensing and communication (ISAC). A novel two-phase sensing framework is proposed to localize the passive targets that cannot transmit/receive reference signals to/from the base stations (BSs), where the ranges of the targets are estimated based on their reflected OFDM signals to the BSs in Phase I, and the location of each target is estimated based on its values of distance to the BSs in Phase II. Specifically, in Phase I, we design a model-free range estimation approach by leveraging the OFDM channel estimation technique for determining the delay values of all the BS-target-BS paths, which does not rely on any BS-target channel model. In Phase II, we reveal that ghost targets may be falsely detected in some cases as all the targets reflect the same signals to the BSs, which thus do not know how to match each estimated range with the right target. Interestingly, we show that the above issue is not a fundamental limitation for device-free sensing: under the ideal case of perfect range estimation in Phase I, the probability for ghost targets to exist is proved to be negligible when the targets are randomly located. Moreover, under the practical case of imperfect range estimation, we propose an efficient algorithm for joint range matching and target localization. Numerical results show that our proposed framework can achieve very high accuracy in the localization of passive targets, which increases with the system bandwidth.
Medical imaging technologies, including computed tomography (CT) or chest X-Ray (CXR), are largely employed to facilitate the diagnosis of the COVID-19. Since manual report writing is usually too time-consuming, a more intelligent auxiliary medical system that could generate medical reports automatically and immediately is urgently needed. In this article, we propose to use the medical visual language BERT (Medical-VLBERT) model to identify the abnormality on the COVID-19 scans and generate the medical report automatically based on the detected lesion regions. To produce more accurate medical reports and minimize the visual-and-linguistic differences, this model adopts an alternate learning strategy with two procedures that are knowledge pretraining and transferring. To be more precise, the knowledge pretraining procedure is to memorize the knowledge from medical texts, while the transferring procedure is to utilize the acquired knowledge for professional medical sentences generations through observations of medical images. In practice, for automatic medical report generation on the COVID-19 cases, we constructed a dataset of 368 medical findings in Chinese and 1104 chest CT scans from The First Affiliated Hospital of Jinan University, Guangzhou, China, and The Fifth Affiliated Hospital of Sun Yat-sen University, Zhuhai, China. Besides, to alleviate the insufficiency of the COVID-19 training samples, our model was first trained on the large-scale Chinese CX-CHR dataset and then transferred to the COVID-19 CT dataset for further fine-tuning. The experimental results showed that Medical-VLBERT achieved state-of-the-art performances on terminology prediction and report generation with the Chinese COVID-19 CT dataset and the CX-CHR dataset. The Chinese COVID-19 CT dataset is available at https://covid19ct.github.io/.
Current 3D single object tracking approaches track the target based on a feature comparison between the target template and the search area. However, due to the common occlusion in LiDAR scans, it is non-trivial to conduct accurate feature comparisons on severe sparse and incomplete shapes. In this work, we exploit the ground truth bounding box given in the first frame as a strong cue to enhance the feature description of the target object, enabling a more accurate feature comparison in a simple yet effective way. In particular, we first propose the BoxCloud, an informative and robust representation, to depict an object using the point-to-box relation. We further design an efficient box-aware feature fusion module, which leverages the aforementioned BoxCloud for reliable feature matching and embedding. Integrating the proposed general components into an existing model P2B, we construct a superior box-aware tracker (BAT). Experiments confirm that our proposed BAT outperforms the previous state-of-the-art by a large margin on both KITTI and NuScenes benchmarks, achieving a 12.8% improvement in terms of precision while running ~20% faster.
Differentiation of colorectal polyps is an important clinical examination. A computer-aided diagnosis system is required to assist accurate diagnosis from colonoscopy images. Most previous studies at-tempt to develop models for polyp differentiation using Narrow-Band Imaging (NBI) or other enhanced images. However, the wide range of these models' applications for clinical work has been limited by the lagging of imaging techniques. Thus, we propose a novel framework based on a teacher-student architecture for the accurate colorectal polyp classification (CPC) through directly using white-light (WL) colonoscopy images in the examination. In practice, during training, the auxiliary NBI images are utilized to train a teacher network and guide the student network to acquire richer feature representation from WL images. The feature transfer is realized by domain alignment and contrastive learning. Eventually the final student network has the ability to extract aligned features from only WL images to facilitate the CPC task. Besides, we release the first public-available paired CPC dataset containing WL-NBI pairs for the alignment training. Quantitative and qualitative evaluation indicates that the proposed method outperforms the previous methods in CPC, improving the accuracy by 5.6%with very fast speed.