Yiran Shen

Kai Bian, Xucheng Guo, Bin Chen et al.

Evaluations of dental vision-language models remain fragmented across datasets, task definitions and metrics, and often ignore their computational cost. This limits their widespread deployment for dental screening outside specialist centres, where timely inference, limited hardware, and local handling of patient images are vital for practical, privacy-preserving clinical prescreening. Here we present Pocket-Dentist, an efficiency-aware benchmark for dental multimodal question answering that brings together three datasets spanning approximately 1,159 patients, five task types and seven metrics. Across typical 14 VLMs, our results reveals an interesting observation: compact VLMs (e.g., 2B-parameter models) outperform larger VLMs in accuracy while requiring substantially lower computational costs in dental image understanding. Deployed locally on an iPhone 17 Pro, our finetuned compact VLM Pocket-Dentist-2B processed each sample in 4.31 s, reducing latency by 4.9-fold and memory use by 2.3-fold compared with a 7B baseline.

Guangrong Zhao, Yiran Shen, Ning Chen et al.

Rotational speed is one of the important metrics to be measured for calibrating the electric motors in manufacturing, monitoring engine during car repairing, faults detection on electrical appliance and etc. However, existing measurement techniques either require prohibitive hardware (e.g., high-speed camera) or are inconvenient to use in real-world application scenarios. In this paper, we propose, EV-Tach, an event-based tachometer via efficient dynamic vision sensing on mobile devices. EV-Tach is designed as a high-fidelity and convenient tachometer by introducing dynamic vision sensor as a new sensing modality to capture the high-speed rotation precisely under various real-world scenarios. By designing a series of signal processing algorithms bespoke for dynamic vision sensing on mobile devices, EV-Tach is able to extract the rotational speed accurately from the event stream produced by dynamic vision sensing on rotary targets. According to our extensive evaluations, the Relative Mean Absolute Error (RMAE) of EV-Tach is as low as 0.03% which is comparable to the state-of-the-art laser tachometer under fixed measurement mode. Moreover, EV-Tach is robust to subtle movement of user's hand, therefore, can be used as a handheld device, where the laser tachometer fails to produce reasonable results.

Yunfan Lu, Guoqiang Liang, Yiran Shen et al.

Most consumer cameras use rolling shutter (RS) exposure, which often leads to distortions such as skew and jelly effects. These videos are further limited by bandwidth and frame rate constraints. In this paper, we explore the potential of event cameras, which offer high temporal resolution. We propose a framework to recover global shutter (GS) high-frame-rate videos without RS distortion by combining an RS camera and an event camera. Due to the lack of real-world datasets, our framework adopts a self-supervised strategy based on a displacement field, a dense 3D spatiotemporal representation of pixel motion during exposure. This enables mutual reconstruction between RS and GS frames and facilitates slow-motion recovery. We combine RS frames with the displacement field to generate GS frames, and integrate inverse mapping and RS frame warping for self-supervision. Experiments on four datasets show that our method removes distortion, reduces bandwidth usage by 94 percent, and achieves 16 ms per frame at 32x interpolation.

Ziyu Luo, Lin Chen, Qiang Qu et al.

Spatial audio is crucial for immersive 360-degree video experiences, yet most 360-degree videos lack it due to the difficulty of capturing spatial audio during recording. Automatically generating spatial audio such as first-order ambisonics (FOA) from video therefore remains an important but challenging problem. In complex scenes, sound perception depends not only on sound source locations but also on scene geometry, materials, and dynamic interactions with the environment. However, existing approaches only rely on visual cues and fail to model dynamic sources and acoustic effects such as occlusion, reflections, and reverberation. To address these challenges, we propose DynFOA, a generative framework that synthesizes FOA from 360-degree videos by integrating dynamic scene reconstruction with conditional diffusion modeling. DynFOA analyzes the input video to detect and localize dynamic sound sources, estimate depth and semantics, and reconstruct scene geometry and materials using 3D Gaussian Splatting (3DGS). The reconstructed scene representation provides physically grounded features that capture acoustic interactions between sources, environment, and listener viewpoint. Conditioned on these features, a diffusion model generates spatial audio consistent with the scene dynamics and acoustic context. We introduce M2G-360, a dataset of 600 real-world clips divided into MoveSources, Multi-Source, and Geometry subsets for evaluating robustness under diverse conditions. Experiments show that DynFOA consistently outperforms existing methods in spatial accuracy, acoustic fidelity, distribution matching, and perceived immersive experience.

Zeke Zexi Hu, Haodong Chen, Hui Ye et al.

A light field image captures scenes through its micro-lens array, providing a rich representation that encompasses spatial and angular information. While this richness comes at significant data redundancy, most existing methods tend to indiscriminately utilize all the information from sub-aperture images (SAIs) in an attempt to harness every visual cue regardless of their disparity significance. However, this paradigm inevitably leads to disparity entanglement, a fundamental cause of inefficiency in light field image processing. To address this limitation, we introduce the Skim Transformer, a novel architecture inspired by the "less is more" philosophy. It features a multi-branch structure where each branch is dedicated to a specific disparity range by constructing its attention score matrix over a skimmed subset of SAIs, rather than all of them. Building upon it, we present SkimLFSR, an efficient yet powerful network for light field image super-resolution. Requiring only 67% of the prior leading method's parameters}, SkimLFSR achieves state-of-the-art results surpassing the best existing method by 0.63 dB and 0.35 dB PSNR at the 2x and 4x tasks, respectively. Through in-depth analyses, we reveal that SkimLFSR, guided by the predefined skimmed SAI sets as prior knowledge, demonstrates distinct disparity-aware behaviors in attending to visual cues. Last but not least, we conduct an experiment to validate SkimLFSR's generalizability across different angular resolutions, where it achieves competitive performance on a larger angular resolution without any retraining or major network modifications. These findings highlight its effectiveness and adaptability as a promising paradigm for light field image processing.

Qiang Qu, Hanxue Liang, Xiaoming Chen et al.

Neural View Synthesis (NVS) has demonstrated efficacy in generating high-fidelity dense viewpoint videos using a image set with sparse views. However, existing quality assessment methods like PSNR, SSIM, and LPIPS are not tailored for the scenes with dense viewpoints synthesized by NVS and NeRF variants, thus, they often fall short in capturing the perceptual quality, including spatial and angular aspects of NVS-synthesized scenes. Furthermore, the lack of dense ground truth views makes the full reference quality assessment on NVS-synthesized scenes challenging. For instance, datasets such as LLFF provide only sparse images, insufficient for complete full-reference assessments. To address the issues above, we propose NeRF-NQA, the first no-reference quality assessment method for densely-observed scenes synthesized from the NVS and NeRF variants. NeRF-NQA employs a joint quality assessment strategy, integrating both viewwise and pointwise approaches, to evaluate the quality of NVS-generated scenes. The viewwise approach assesses the spatial quality of each individual synthesized view and the overall inter-views consistency, while the pointwise approach focuses on the angular qualities of scene surface points and their compound inter-point quality. Extensive evaluations are conducted to compare NeRF-NQA with 23 mainstream visual quality assessment methods (from fields of image, video, and light-field assessment). The results demonstrate NeRF-NQA outperforms the existing assessment methods significantly and it shows substantial superiority on assessing NVS-synthesized scenes without references. An implementation of this paper are available at https://github.com/VincentQQu/NeRF-NQA.

Ziyu Luo, Lin Chen, Qiang Qu et al.

Spatial audio is crucial for immersive 360-degree video experiences, yet most 360-degree videos lack it due to the difficulty of capturing spatial audio during recording. Automatically generating spatial audio such as first-order ambisonics (FOA) from video therefore remains an important but challenging problem. In complex scenes, sound perception depends not only on sound source locations but also on scene geometry, materials, and dynamic interactions with the environment. However, existing approaches only rely on visual cues and fail to model dynamic sources and acoustic effects such as occlusion, reflections, and reverberation. To address these challenges, we propose DynFOA, a generative framework that synthesizes FOA from 360-degree videos by integrating dynamic scene reconstruction with conditional diffusion modeling. DynFOA analyzes the input video to detect and localize dynamic sound sources, estimate depth and semantics, and reconstruct scene geometry and materials using 3D Gaussian Splatting (3DGS). The reconstructed scene representation provides physically grounded features that capture acoustic interactions between sources, environment, and listener viewpoint. Conditioned on these features, a diffusion model generates spatial audio consistent with the scene dynamics and acoustic context. We introduce M2G-360, a dataset of 600 real-world clips divided into MoveSources, Multi-Source, and Geometry subsets for evaluating robustness under diverse conditions. Experiments show that DynFOA consistently outperforms existing methods in spatial accuracy, acoustic fidelity, distribution matching, and perceived immersive experience.

Kaiwen Xue, Chenglong Li, Zhonghong Ou et al.

Human-defined creativity is highly abstract, posing a challenge for multimodal large language models (MLLMs) to comprehend and assess creativity that aligns with human judgments. The absence of an existing benchmark further exacerbates this dilemma. To this end, we propose CreBench, which consists of two key components: 1) an evaluation benchmark covering the multiple dimensions from creative idea to process to products; 2) CreMIT (Creativity Multimodal Instruction Tuning dataset), a multimodal creativity evaluation dataset, consisting of 2.2K diverse-sourced multimodal data, 79.2K human feedbacks and 4.7M multi-typed instructions. Specifically, to ensure MLLMs can handle diverse creativity-related queries, we prompt GPT to refine these human feedbacks to activate stronger creativity assessment capabilities. CreBench serves as a foundation for building MLLMs that understand human-aligned creativity. Based on the CreBench, we fine-tune open-source general MLLMs, resulting in CreExpert, a multimodal creativity evaluation expert model. Extensive experiments demonstrate that the proposed CreExpert models achieve significantly better alignment with human creativity evaluation compared to state-of-the-art MLLMs, including the most advanced GPT-4V and Gemini-Pro-Vision.

Yiran Shen, Yu Xia, Jonathan Chang et al.

Aligning large language models to human preferences is inherently multidimensional, yet most pipelines collapse heterogeneous signals into a single optimizeable objective. We seek to answer what it would take to simultaneously align a model across various domains spanning those with: verifiable rewards (mathematical accuracy), non-verifiable subjective preferences (human values), and complex interactive scenarios (multi-turn AI tutoring dialogues). Such multi-objective reinforcement learning setups are often plagued by the individual objectives being at odds with each other, resulting in inefficient training and little user control during inference. We propose a unified framework that: (i) standardizes {process reward model} (PRM) training across both verifiable and non-verifiable settings to better supervise models' chain-of-thought reasoning; (ii) performs {multi-objective alignment} by training the LLM with our $\textbf{M}$ulti-$\textbf{A}$ction-$\textbf{H}$ead $\textbf{DPO}$ (MAH-DPO) and a vectorized reward where the dimensions of the vector correspond to the various objectives instead of a single scalar; and (iii) demonstrates how such a system provides fine-grained inference-time user control. Experiments across math reasoning, value alignment, and multi-turn dialogue show that our framework improves performance across multiple objectives simultaneously, while minimizing cross-objective trade-offs and enabling flexible inference time user control. The code can be found at https://github.com/pearls-lab/multiobj-align.

Xiaofang Xiao, Guangchao Li, Guangrong Zhao et al.

Automatic sign language recognition (SLR) has become a key enabler of inclusive human-computer interaction, fostering seamless communication between deaf individuals and hearing communities. Despite significant advances in multimodal learning, existing SLR research remains dominated by vision-based datasets, which are limited by sensitivity to lighting and occlusion, privacy concerns, and a lack of cross-modal diversity. To address these challenges, we introduce SIGMA-ASL, a large-scale multimodal dataset for SLR. The dataset integrates an Azure Kinect RGB-D camera, a millimeter-wave (mmWave) radar, and two wrist-worn inertial measurement units (IMUs) to capture complementary visual, radio-reflection, and kinematic information. Collected in a controlled studio environment with 20 participants performing 160 common American sign language (ASL) signs, SIGMA-ASL provides 93,545 temporally synchronized word-level multimodal clips. A unified sensing framework achieves millisecond-level alignment across modalities, enabling reliable sensor fusion and cross-modal learning. We further design standardized preprocessing pipelines and benchmarking protocols under both user-dependent and user-independent settings, offering a comprehensive foundation for evaluating single and multimodal SLR. Extensive experiments validate the dataset's quality and demonstrate its potential as a valuable resource for developing robust, privacy-preserving, and ubiquitous sign language recognition systems.

Qiang Qu, Xiaoming Chen, Yuk Ying Chung et al.

Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representations are manually designed and the quality of these representations cannot be guaranteed due to the noisy nature of event-streams. In this paper, we introduce a data-driven approach aiming at enhancing the quality of event-stream representations. Our approach commences with the introduction of a new event-stream representation based on spatial-temporal statistics, denoted as EvRep. Subsequently, we theoretically derive the intrinsic relationship between asynchronous event-streams and synchronous video frames. Building upon this theoretical relationship, we train a representation generator, RepGen, in a self-supervised learning manner accepting EvRep as input. Finally, the event-streams are converted to high-quality representations, termed as EvRepSL, by going through the learned RepGen (without the need of fine-tuning or retraining). Our methodology is rigorously validated through extensive evaluations on a variety of mainstream event-based classification and optical flow datasets (captured with various types of event cameras). The experimental results highlight not only our approach's superior performance over existing event-stream representations but also its versatility, being agnostic to different event cameras and tasks.

Zhouhang Xie, Junda Wu, Yiran Shen et al.

Personalized preference alignment for large language models (LLMs), the process of tailoring LLMs to individual users' preferences, is an emerging research direction spanning the area of NLP and personalization. In this survey, we present an analysis of works on personalized alignment and modeling for LLMs. We introduce a taxonomy of preference alignment techniques, including training time, inference time, and additionally, user-modeling based methods. We provide analysis and discussion on the strengths and limitations of each group of techniques and then cover evaluation, benchmarks, as well as open problems in the field.

Zeke Zexi Hu, Xiaoming Chen, Vera Yuk Ying Chung et al.

The effective extraction of spatial-angular features plays a crucial role in light field image super-resolution (LFSR) tasks, and the introduction of convolution and Transformers leads to significant improvement in this area. Nevertheless, due to the large 4D data volume of light field images, many existing methods opted to decompose the data into a number of lower-dimensional subspaces and perform Transformers in each sub-space individually. As a side effect, these methods inadvertently restrict the self-attention mechanisms to a One-to-One scheme accessing only a limited subset of LF data, explicitly preventing comprehensive optimization on all spatial and angular cues. In this paper, we identify this limitation as subspace isolation and introduce a novel Many-to-Many Transformer (M2MT) to address it. M2MT aggregates angular information in the spatial subspace before performing the self-attention mechanism. It enables complete access to all information across all sub-aperture images (SAIs) in a light field image. Consequently, M2MT is enabled to comprehensively capture long-range correlation dependencies. With M2MT as the foundational component, we develop a simple yet effective M2MT network for LFSR. Our experimental results demonstrate that M2MT achieves state-of-the-art performance across various public datasets, and it offers a favorable balance between model performance and efficiency, yielding higher-quality LFSR results with substantially lower demand for memory and computation. We further conduct in-depth analysis using local attribution maps (LAM) to obtain visual interpretability, and the results validate that M2MT is empowered with a truly non-local context in both spatial and angular subspaces to mitigate subspace isolation and acquire effective spatial-angular representation.

Junda Wu, Rohan Surana, Zhouhang Xie et al.

Recent developments in Direct Preference Optimization (DPO) allow large language models (LLMs) to function as implicit ranking models by maximizing the margin between preferred and non-preferred responses. In practice, user feedback on such lists typically involves identifying a few relevant items in context rather than providing detailed pairwise comparisons for every possible item pair. Moreover, many complex information retrieval tasks, such as conversational agents and summarization systems, critically depend on ranking the highest-quality outputs at the top, emphasizing the need to support natural and flexible forms of user feedback. To address the challenge of limited and sparse pairwise feedback in the in-context setting, we propose an In-context Ranking Preference Optimization (IRPO) framework that directly optimizes LLMs based on ranking lists constructed during inference. To further capture flexible forms of feedback, IRPO extends the DPO objective by incorporating both the relevance of items and their positions in the list. Modeling these aspects jointly is non-trivial, as ranking metrics are inherently discrete and non-differentiable, making direct optimization difficult. To overcome this, IRPO introduces a differentiable objective based on positional aggregation of pairwise item preferences, enabling effective gradient-based optimization of discrete ranking metrics. We further provide theoretical insights showing that IRPO (i) automatically emphasizes items with greater disagreement between the model and the reference ranking, and (ii) links its gradient to an importance sampling estimator, yielding an unbiased estimator with reduced variance. Empirical results show IRPO outperforms standard DPO approaches in ranking performance, highlighting its effectiveness in aligning LLMs with direct in-context ranking preferences.

Yu Xia, Yiran Shen, Junda Wu et al.

Large Language Model (LLM) agents are commonly tuned with supervised finetuning on ReAct-style expert trajectories or preference optimization over pairwise rollouts. Most of these methods focus on imitating specific expert behaviors or promoting chosen reasoning thoughts and actions over rejected ones. However, without reasoning and comparing over alternatives actions, LLM agents finetuned with these methods may over-commit towards seemingly plausible but suboptimal actions due to limited action space exploration. To address this, in this paper we propose Self-taught ActioN Deliberation (SAND) framework, enabling LLM agents to explicitly deliberate over candidate actions before committing to one. To tackle the challenges of when and what to deliberate given large action space and step-level action evaluation, we incorporate self-consistency action sampling and execution-guided action critique to help synthesize step-wise action deliberation thoughts using the base model of the LLM agent. In an iterative manner, the deliberation trajectories are then used to finetune the LLM agent itself. Evaluating on two representative interactive agent tasks, SAND achieves an average 20% improvement over initial supervised finetuning and also outperforms state-of-the-art agent tuning approaches.

Qiang Qu, Yiran Shen, Xiaoming Chen et al.

Neural View Synthesis (NVS), such as NeRF and 3D Gaussian Splatting, effectively creates photorealistic scenes from sparse viewpoints, typically evaluated by quality assessment methods like PSNR, SSIM, and LPIPS. However, these full-reference methods, which compare synthesized views to reference views, may not fully capture the perceptual quality of neurally synthesized scenes (NSS), particularly due to the limited availability of dense reference views. Furthermore, the challenges in acquiring human perceptual labels hinder the creation of extensive labeled datasets, risking model overfitting and reduced generalizability. To address these issues, we propose NVS-SQA, a NSS quality assessment method to learn no-reference quality representations through self-supervision without reliance on human labels. Traditional self-supervised learning predominantly relies on the "same instance, similar representation" assumption and extensive datasets. However, given that these conditions do not apply in NSS quality assessment, we employ heuristic cues and quality scores as learning objectives, along with a specialized contrastive pair preparation process to improve the effectiveness and efficiency of learning. The results show that NVS-SQA outperforms 17 no-reference methods by a large margin (i.e., on average 109.5% in SRCC, 98.6% in PLCC, and 91.5% in KRCC over the second best) and even exceeds 16 full-reference methods across all evaluation metrics (i.e., 22.9% in SRCC, 19.1% in PLCC, and 18.6% in KRCC over the second best).

Aaron Broukhim, Yiran Shen, Prithviraj Ammanabrolu et al.

Despite the parallel challenges that audio and text domains face in evaluating generative model outputs, preference learning remains remarkably underexplored in audio applications. Through a PRISMA-guided systematic review of approximately 500 papers, we find that only 30 (6%) apply preference learning to audio tasks. Our analysis reveals a field in transition: pre-2021 works focused on emotion recognition using traditional ranking methods (rankSVM), while post-2021 studies have pivoted toward generation tasks employing modern RLHF frameworks. We identify three critical patterns: (1) the emergence of multi-dimensional evaluation strategies combining synthetic, automated, and human preferences; (2) inconsistent alignment between traditional metrics (WER, PESQ) and human judgments across different contexts; and (3) convergence on multi-stage training pipelines that combine reward signals. Our findings suggest that while preference learning shows promise for audio, particularly in capturing subjective qualities like naturalness and musicality, the field requires standardized benchmarks, higher-quality datasets, and systematic investigation of how temporal factors unique to audio impact preference learning frameworks.

Xucheng Guo, Yiran Shen, Xiaofang Xiao et al.

This paper presents Ev-Layout, a novel large-scale event-based multi-modal dataset designed for indoor layout estimation and tracking. Ev-Layout makes key contributions to the community by: Utilizing a hybrid data collection platform (with a head-mounted display and VR interface) that integrates both RGB and bio-inspired event cameras to capture indoor layouts in motion. Incorporating time-series data from inertial measurement units (IMUs) and ambient lighting conditions recorded during data collection to highlight the potential impact of motion speed and lighting on layout estimation accuracy. The dataset consists of 2.5K sequences, including over 771.3K RGB images and 10 billion event data points. Of these, 39K images are annotated with indoor layouts, enabling research in both event-based and video-based indoor layout estimation. Based on the dataset, we propose an event-based layout estimation pipeline with a novel event-temporal distribution feature module to effectively aggregate the spatio-temporal information from events. Additionally, we introduce a spatio-temporal feature fusion module that can be easily integrated into a transformer module for fusion purposes. Finally, we conduct benchmarking and extensive experiments on the Ev-Layout dataset, demonstrating that our approach significantly improves the accuracy of dynamic indoor layout estimation compared to existing event-based methods.

Qiang Qu, Yiran Shen, Xiaoming Chen et al.

The bio-inspired event cameras or dynamic vision sensors are capable of asynchronously capturing per-pixel brightness changes (called event-streams) in high temporal resolution and high dynamic range. However, the non-structural spatial-temporal event-streams make it challenging for providing intuitive visualization with rich semantic information for human vision. It calls for events-to-video (E2V) solutions which take event-streams as input and generate high quality video frames for intuitive visualization. However, current solutions are predominantly data-driven without considering the prior knowledge of the underlying statistics relating event-streams and video frames. It highly relies on the non-linearity and generalization capability of the deep neural networks, thus, is struggling on reconstructing detailed textures when the scenes are complex. In this work, we propose \textbf{E2HQV}, a novel E2V paradigm designed to produce high-quality video frames from events. This approach leverages a model-aided deep learning framework, underpinned by a theory-inspired E2V model, which is meticulously derived from the fundamental imaging principles of event cameras. To deal with the issue of state-reset in the recurrent components of E2HQV, we also design a temporal shift embedding module to further improve the quality of the video frames. Comprehensive evaluations on the real world event camera datasets validate our approach, with E2HQV, notably outperforming state-of-the-art approaches, e.g., surpassing the second best by over 40\% for some evaluation metrics.

Chris Xiaoxuan Lu, Bowen Du, Hongkai Wen et al.

Demand for smartwatches has taken off in recent years with new models which can run independently from smartphones and provide more useful features, becoming first-class mobile platforms. One can access online banking or even make payments on a smartwatch without a paired phone. This makes smartwatches more attractive and vulnerable to malicious attacks, which to date have been largely overlooked. In this paper, we demonstrate Snoopy, a password extraction and inference system which is able to accurately infer passwords entered on Android/Apple watches within 20 attempts, just by eavesdropping on motion sensors. Snoopy uses a uniform framework to extract the segments of motion data when passwords are entered, and uses novel deep neural networks to infer the actual passwords. We evaluate the proposed Snoopy system in the real-world with data from 362 participants and show that our system offers a 3-fold improvement in the accuracy of inferring passwords compared to the state-of-the-art, without consuming excessive energy or computational resources. We also show that Snoopy is very resilient to user and device heterogeneity: it can be trained on crowd-sourced motion data (e.g. via Amazon Mechanical Turk), and then used to attack passwords from a new user, even if they are wearing a different model. This paper shows that, in the wrong hands, Snoopy can potentially cause serious leaks of sensitive information. By raising awareness, we invite the community and manufacturers to revisit the risks of continuous motion sensing on smart wearable devices.

Yiran Shen, Fengyuan Yang, Bowen Du et al.

Since ancient Greece, handshaking has been commonly practiced between two people as a friendly gesture to express trust and respect, or form a mutual agreement. In this paper, we show that such physical contact can be used to bootstrap secure cyber contact between the smart devices worn by users. The key observation is that during handshaking, although belonged to two different users, the two hands involved in the shaking events are often rigidly connected, and therefore exhibit very similar motion patterns. We propose a novel Shake-n-Shack system, which harvests motion data during user handshaking from the wrist worn smart devices such as smartwatches or fitness bands, and exploits the matching motion patterns to generate symmetric keys on both parties. The generated keys can be then used to establish a secure communication channel for exchanging data between devices. This provides a much more natural and user-friendly alternative for many applications, e.g. exchanging/sharing contact details, friending on social networks, or even making payments, since it doesn't involve extra bespoke hardware, nor require the users to perform pre-defined gestures. We implement the proposed Shake-n-Shack system on off-the-shelf smartwatches, and extensive evaluation shows that it can reliably generate 128-bit symmetric keys just after around 1s of handshaking (with success rate >99%), and is resilient to real-time mimicking attacks: in our experiments the Equal Error Rate (EER) is only 1.6% on average. We also show that the proposed Shake-n-Shack system can be extremely lightweight, and is able to run in-situ on the resource-constrained smartwatches without incurring excessive resource consumption.