arXiv subjects

Yuanpeng Liu

Publications and source records attributed to Yuanpeng Liu.

17 recordsLinked to original sources

arXiv2026

A General-Purpose VLM Can Teach an Astronomy Foundation Model to Better Recognize Galaxy Morphology

Existing astronomy foundation models provide strong galaxy representations, but adapting them to new survey conditions and survey-specific morphology recognition tasks still requires substantial human supervision. We show that VLM-based VQA systems contain meaningful visual-semantic priors that can serve as weak supervision for downstream morphology classifiers and improve morphology classification under limited human-label budgets. We first introduce a survey-oriented VQA benchmark spanning two representative imaging regimes and evaluate state-of-the-art VLMs on galaxy morphology questions. The results show that these models capture useful morphology signals and informative uncertainty, but are not sufficiently reliable to replace human annotators. Motivated by this finding, we use a general-purpose VLM as a morphology teacher for Zoobot, an astronomy foundation model pretrained on large-scale Galaxy Zoo annotations. Across two survey domains and multiple annotation budgets, the VLM teacher consistently improves Zoobot's downstream morphology classification. These results demonstrate that a general-purpose VLM provides knowledge complementary to an astronomy foundation model and can teach it to better recognize galaxy morphology under limited human supervision. The resulting pipeline is designed for label-efficient adaptation to forthcoming large-scale surveys, including the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) and the Nancy Grace Roman Space Telescope. The benchmark and code are publicly available at https://github.com/fw-ic/VLM-morphology-teacher.

cs.CV↗

arXiv2026

Inverse Rendering for High-Genus 3D Surface Meshes from Multi-view Images with Persistent Homology Priors

Reconstructing 3D objects from images is inherently an ill-posed problem due to ambiguities in geometry, appearance, and topology. This paper introduces collaborative inverse rendering with persistent homology priors, a novel strategy that leverages topological constraints to resolve these ambiguities. By incorporating priors that capture critical features such as tunnel loops and handle loops, our approach directly addresses the difficulty of reconstructing high-genus surfaces. The collaboration between photometric consistency from multi-view images and homology-based guidance enables recovery of complex high-genus geometry while circumventing catastrophic failures such as collapsing tunnels or losing high-genus structure. Instead of neural networks, our method relies on gradient-based optimization within a mesh-based inverse rendering framework to highlight the role of topological priors. Experimental results show that incorporating persistent homology priors leads to lower Chamfer Distance (CD) and higher Volume IoU compared to state-of-the-art mesh-based methods, demonstrating improved geometric accuracy and robustness against topological failure.

cs.CV↗

arXiv2026

Learning Domain Agnostic Latent Embeddings of 3D Faces for Zero-shot Animal Expression Transfer

We present a zero-shot framework for transferring human facial expressions to 3D animal face meshes. Our method combines intrinsic geometric descriptors (HKS/WKS) with a mesh-agnostic latent embedding that disentangles facial identity and expression. The ID latent space captures species-independent facial structure, while the expression latent space encodes deformation patterns that generalize across humans and animals. Trained only with human expression pairs, the model learns the embeddings, decoupling, and recoupling of cross-identity expressions, enabling expression transfer without requiring animal expression data. To enforce geometric consistency, we employ Jacobian loss together with vertex-position and Laplacian losses. Experiments show that our approach achieves plausible cross-species expression transfer, effectively narrowing the geometric gap between human and animal facial shapes.

cs.CV↗

arXiv2025

Neural Geometry Image-Based Representations with Optimal Transport (OT)

Neural representations for 3D meshes are emerging as an effective solution for compact storage and efficient processing. Existing methods often rely on neural overfitting, where a coarse mesh is stored and progressively refined through multiple decoder networks. While this can restore high-quality surfaces, it is computationally expensive due to successive decoding passes and the irregular structure of mesh data. In contrast, images have a regular structure that enables powerful super-resolution and restoration frameworks, but applying these advantages to meshes is difficult because their irregular connectivity demands complex encoder-decoder architectures. Our key insight is that a geometry image-based representation transforms irregular meshes into a regular image grid, making efficient image-based neural processing directly applicable. Building on this idea, we introduce our neural geometry image-based representation, which is decoder-free, storage-efficient, and naturally suited for neural processing. It stores a low-resolution geometry-image mipmap of the surface, from which high-quality meshes are restored in a single forward pass. To construct geometry images, we leverage Optimal Transport (OT), which resolves oversampling in flat regions and undersampling in feature-rich regions, and enables continuous levels of detail (LoD) through geometry-image mipmapping. Experimental results demonstrate state-of-the-art storage efficiency and restoration accuracy, measured by compression ratio (CR), Chamfer distance (CD), and Hausdorff distance (HD).

cs.CV↗

arXiv2025

Inverse Rendering for High-Genus Surface Meshes from Multi-View Images

We present a topology-informed inverse rendering approach for reconstructing high-genus surface meshes from multi-view images. Compared to 3D representations like voxels and point clouds, mesh-based representations are preferred as they enable the application of differential geometry theory and are optimized for modern graphics pipelines. However, existing inverse rendering methods often fail catastrophically on high-genus surfaces, leading to the loss of key topological features, and tend to oversmooth low-genus surfaces, resulting in the loss of surface details. This failure stems from their overreliance on Adam-based optimizers, which can lead to vanishing and exploding gradients. To overcome these challenges, we introduce an adaptive V-cycle remeshing scheme in conjunction with a re-parametrized Adam optimizer to enhance topological and geometric awareness. By periodically coarsening and refining the deforming mesh, our method informs mesh vertices of their current topology and geometry before optimization, mitigating gradient issues while preserving essential topological features. Additionally, we enforce topological consistency by constructing topological primitives with genus numbers that match those of ground truth using Gauss-Bonnet theorem. Experimental results demonstrate that our inverse rendering approach outperforms the current state-of-the-art method, achieving significant improvements in Chamfer Distance and Volume IoU, particularly for high-genus surfaces, while also enhancing surface details for low-genus surfaces.

cs.GR↗

arXiv2022

SO(3)-Pose: SO(3)-Equivariance Learning for 6D Object Pose Estimation

6D pose estimation of rigid objects from RGB-D images is crucial for object grasping and manipulation in robotics. Although RGB channels and the depth (D) channel are often complementary, providing respectively the appearance and geometry information, it is still non-trivial how to fully benefit from the two cross-modal data. From the simple yet new observation, when an object rotates, its semantic label is invariant to the pose while its keypoint offset direction is variant to the pose. To this end, we present SO(3)-Pose, a new representation learning network to explore SO(3)-equivariant and SO(3)-invariant features from the depth channel for pose estimation. The SO(3)-invariant features facilitate to learn more distinctive representations for segmenting objects with similar appearance from RGB channels. The SO(3)-equivariant features communicate with RGB features to deduce the (missed) geometry for detecting keypoints of an object with the reflective surface from the depth channel. Unlike most of existing pose estimation methods, our SO(3)-Pose not only implements the information communication between the RGB and depth channels, but also naturally absorbs the SO(3)-equivariance geometry knowledge from depth images, leading to better appearance and geometry representation learning. Comprehensive experiments show that our method achieves the state-of-the-art performance on three benchmarks.

cs.CV↗

arXiv2021

Automatic CT Segmentation from Bounding Box Annotations using Convolutional Neural Networks

Accurate segmentation for medical images is important for clinical diagnosis. Existing automatic segmentation methods are mainly based on fully supervised learning and have an extremely high demand for precise annotations, which are very costly and time-consuming to obtain. To address this problem, we proposed an automatic CT segmentation method based on weakly supervised learning, by which one could train an accurate segmentation model only with weak annotations in the form of bounding boxes. The proposed method is composed of two steps: 1) generating pseudo masks with bounding box annotations by k-means clustering, and 2) iteratively training a 3D U-Net convolutional neural network as a segmentation model. Some data pre-processing methods are used to improve performance. The method was validated on four datasets containing three types of organs with a total of 627 CT volumes. For liver, spleen and kidney segmentation, it achieved an accuracy of 95.19%, 92.11%, and 91.45%, respectively. Experimental results demonstrate that our method is accurate, efficient, and suitable for clinical use.

cs.CV↗

arXiv2021

Deep learning for prediction of hepatocellular carcinoma recurrence after resection or liver transplantation: a discovery and validation study

This study aimed to develop a classifier of prognosis after resection or liver transplantation (LT) for HCC by directly analysing the ubiquitously available histological images using deep learning based neural networks. Nucleus map set was used to train U-net to capture the nuclear architectural information. Train set included the patients with HCC treated by resection and has a distinct outcome. LT set contained patients with HCC treated by LT. Train set and its nuclear architectural information extracted by U-net were used to train MobileNet V2 based classifier (MobileNetV2_HCC_Class), purpose-built for classifying supersized heterogeneous images. The MobileNetV2_HCC_Class maintained relative higher discriminatory power than the other factors after HCC resection or LT in the independent validation set. Pathological review showed that the tumoral areas most predictive of recurrence were characterized by presence of stroma, high degree of cytological atypia, nuclear hyperchomasia, and a lack of immune infiltration. A clinically useful prognostic classifier was developed using deep learning allied to histological slides. The classifier has been extensively evaluated in independent patient populations with different treatment, and gives consistent excellent results across the classical clinical, biological and pathological features. The classifier assists in refining the prognostic prediction of HCC patients and identifying patients who would benefit from more intensive management.

cs.CV↗

arXiv2020

3DPVNet: Patch-level 3D Hough Voting Network for 6D Pose Estimation

In this paper, we focus on estimating the 6D pose of objects in point clouds. Although the topic has been widely studied, pose estimation in point clouds remains a challenging problem due to the noise and occlusion. To address the problem, a novel 3DPVNet is presented in this work, which utilizes 3D local patches to vote for the object 6D poses. 3DPVNet is comprised of three modules. In particular, a Patch Unification (\textbf{PU}) module is first introduced to normalize the input patch, and also create a standard local coordinate frame on it to generate a reliable vote. We then devise a Weight-guided Neighboring Feature Fusion (\textbf{WNFF}) module in the network, which fuses the neighboring features to yield a semi-global feature for the center patch. WNFF module mines the neighboring information of a local patch, such that the representation capability to local geometric characteristics is significantly enhanced, making the method robust to a certain level of noise. Moreover, we present a Patch-level Voting (\textbf{PV}) module to regress transformations and generates pose votes. After the aggregation of all votes from patches and a refinement step, the final pose of the object can be obtained. Compared to recent voting-based methods, 3DPVNet is patch-level, and directly carried out on point clouds. Therefore, 3DPVNet achieves less computation than point/pixel-level voting scheme, and has robustness to partial data. Experiments on several datasets demonstrate that 3DPVNet achieves the state-of-the-art performance, and is also robust against noise and occlusions.

cs.CV↗

arXiv2015

Completion Time in Two-user Channels: An Information-Theoretic Perspective

In a two-user channel, completion time refers to the number of channel uses spent by each user to transmit a bit pool with some given size. In this paper, the information-theoretic formulation of completion time is based on the concept of constrained rates, where users are allowed to employ different numbers of channel uses for transmission as opposed to the equal channel use of the standard information-theoretic formulation. Analogous to the capacity region, the completion time region characterizes all possible trade-offs among users' completion times. For a multi-access channel, it is shown that the completion time region is achieved by operating the channel in two independent phases: a multi-access phase when both users are transmitting, and a point-to-point phase when one user has finished and the other is still transmitting. Using a similar two-phase approach, the completion time region (or inner and outer bounds) is established for a Gaussian broadcast channel and a Gaussian interference channel. It is observed that although consisting of two convex subregions, the completion time region may not be convex in general. Finally an optimization problem of minimizing the weighted sum completion time for a Gaussian multi-access channel and a Gaussian broadcast channel is solved, demonstrating the utility of the completion time approach.

cs.IT↗

arXiv2015

Capacity and Rate Regions of A Class of Broadcast Interference Channels

In this paper, a class of broadcast interference channels (BIC) is investigated, where one of the two broadcast receivers is subject to interference coming from a point-to-point transmission. For a general discrete memoryless broadcast interference channel (DM-BIC), an achievable scheme based on message splitting, superposition and binning is proposed and a concise representation of the corresponding achievable rate region R is obtained. Two partial-order broadcast conditions interference-oblivious less noisy and interference-cognizant less noisy are defined, thereby extending the usual less noisy condition for a regular broadcast channel by taking interference into account. Under these conditions, a reduced form of R is shown to be equivalent to a rate region based on a simpler scheme, where the broadcast transmitter uses only superposition. Furthermore, if interference is strong for the interference-oblivious less noisy DM-BIC, the capacity region is given by the aforementioned two equivalent rate regions. For a Gaussian broadcast interference channel (GBIC), channel parameters are categorized into three regimes. For the first two regimes, which are closely related to the two partial-order broadcast conditions, achievable rate regions are derived by specializing the corresponding achievable schemes of DM-BICs with Gaussian input distributions. The entropy power inequality (EPI) based outer bounds are obtained by combining bounding techniques for a Gaussian broadcast channel (GBC) and a Gaussian interference channel (GIC). These inner and outer bounds lead to either exact or approximate characterizations of capacity regions and sum capacity under various conditions. For the remaining complementing regime, inner and outer bounds are also provided.

cs.IT↗

arXiv2014

Millimeter Wave Picocellular System Evaluation for Urban Deployments

With the severe spectrum shortage in conventional cellular bands, millimeter wave (mmW) frequencies between 30 and 300 GHz have been attracting growing attention as a possible candidate for next-generation micro- and picocellular wireless networks. The mmW bands offer orders of magnitude greater spectrum than current cellular allocations and enable very high-dimensional antenna arrays for further gains via spatial multiplexing. However, the propagation of mmW signals in outdoor non line-of-sight (NLOS) links remains challenging and the feasibility of wide-area mmW cellular networks is far from clear. This paper uses recent real-world measurements at 28 GHz in New York City to provide a realistic assessment of mmW picocellular networks in a dense urban deployment. It is found that, even under conservative propagation assumptions, mmW systems with cell radii of 100m can offer an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks with similar cell density. However, it is also shown that such mmW networks may operate in a largely power-limited regime where the full spatial and bandwidth degrees of freedom are not fully utilized. This power-limited regime contrasts significantly with current bandwidth-limited cellular systems, requiring alternate technologies for mmW systems that may unlock further gains that mmW frequency bands offer.

cs.NI↗

arXiv2014

Millimeter Wave Channel Modeling and Cellular Capacity Evaluation

With the severe spectrum shortage in conventional cellular bands, millimeter wave (mmW) frequencies between 30 and 300 GHz have been attracting growing attention as a possible candidate for next-generation micro- and picocellular wireless networks. The mmW bands offer orders of magnitude greater spectrum than current cellular allocations and enable very high-dimensional antenna arrays for further gains via beamforming and spatial multiplexing. This paper uses recent real-world measurements at 28 and 73 GHz in New York City to derive detailed spatial statistical models of the channels and uses these models to provide a realistic assessment of mmW micro- and picocellular networks in a dense urban deployment. Statistical models are derived for key channel parameters including the path loss, number of spatial clusters, angular dispersion and outage. It is found that, even in highly non-line-of-sight environments, strong signals can be detected 100 m to 200 m from potential cell sites, potentially with multiple clusters to support spatial multiplexing. Moreover, a system simulation based on the models predicts that mmW systems can offer an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks with no increase in cell density from current urban deployments.

cs.NI↗

arXiv2012

On a Class of Discrete Memoryless Broadcast Interference Channels

We study a class of discrete memoryless broadcast interference channels (DM-BICs), where one of the broadcast receivers is subject to the interference from a point-to-point transmission. A general achievable rate region $\mathcal{R}$ based on rate splitting, superposition coding and binning at the broadcast transmitter and rate splitting at the interfering transmitter is derived. Under two partial order broadcast conditions {\em interference-oblivious less noisy} and {\em interference-cognizant less noisy}, a reduced form of $\mathcal{R}$ is shown to be equivalent to the region based on a simpler scheme that uses only superposition coding at the broadcast transmitter. Furthermore, the capacity regions of DM-BIC under the two partial order broadcast conditions are characterized respectively for the strong and very strong interference conditions.

cs.IT↗

arXiv2011

Completion Time in Broadcast Channel and Interference Channel

In a multi-user channel, completion time refers to the number of channel uses required for users, each with some given fixed bit pool, to complete the transmission of all their data bits. This paper extends the information theoretic formulation of multi-access completion time to broadcast channel and interference channel, enabling us to obtain the so-called completion time region (CTR), which, analogous to capacity region, characterizes all possible trade-offs between users' completion times. Specifically, for Gaussian broadcast channel (GBC) and Gaussian interference channel (GIC) in the strong/very strong regime, the exact CTR is obtained. For GIC in the weak/mixed regime, an achievable CTR based on the Etkin-Tse-Wang scheme and an outer-bound are obtained.

cs.IT↗

arXiv2011

Completion Time in Multi-Access Channel: An Information Theoretic Perspective

In a multi-access channel, completion time refers to the number of channel uses required for users, each with some given fixed bit pool, to complete the transmission of all their data bits. In this paper, the characterization of the completion time region is based on the concept of constrained rates, where users' rates are defined over possibly different number of channel uses. An information theoretic formulation of completion time is given and the completion time region is then established for two-user Gaussian multi-access channel, which, analogous to capacity region, characterizes all possible trade-offs between users' completion times.

cs.IT↗

arXiv2011

On the Sum Capacity of K-user Cascade Gaussian Z-Interference Channel

A $K$-user cascade Gaussian Z-interference channel is a subclass of the general $K$-user Gaussian interference channel, where each user, except the first one, experiences interference only from the previous user. Under simple Han-Kobayashi schemes assuming Gaussian inputs and no time sharing, it is shown that the maximum sum rate is achieved by each user transmitting either common or private signals. For K=3, channel conditions under which the achieved sum rate is either equal to or within 0.5 bits to the sum capacity are identified.

cs.IT↗