arXiv subjects

Xutong Wang

Publications and source records attributed to Xutong Wang.

6 recordsLinked to original sources

arXiv2026

PAGE: Towards Practical Human-level Gaze Target Estimation

Gaze target estimation, the task of predicting where a person is looking in a scene, is crucial to understanding human attention and intent. It is a challenging task that combines high-level understanding of global scene semantics and precise spatial reasoning using human appearance (e.g. pose, eye orientation). As a result, human-level performance remains elusive for existing models, limiting their practical application. To this end, we propose PaGE (Practical Gaze Estimator), a gaze estimation model that explicitly models the complex interaction between scene and head features. Using a PaGE model with a large ViT-H+ backbone as the teacher, we further distill student models with lighter backbones on a much larger and more diverse unlabeled dataset. The architectural improvements and novel training recipe allow PaGE to achieve state-of-the-art performance on several gaze estimation tasks, outperforming humans in 7 out of 9 metrics while reducing the human-AI gap by at least 60% in the remaining 2. The distilled student models retain most of the teacher's performance while being lightweight enough for practical deployment on robots and consumer devices. The code and model checkpoints are available at our project page.

cs.CV↗

arXiv2026

PAPEL: A Collaborative System for Parental Guidance during Preschool Play-Based English Learning

Play-based parent-child interaction offers preschoolers rich opportunities for everyday foreign language learning, yet many parents struggle to turn open-ended play into effective English-as-a-Foreign-Language (EFL) learning experiences at home. To explore how AI might support this process, we conducted formative studies through interviews and a Wizard-of-Oz study. We identified four key challenges: content selection, language expression, balancing instruction and play, and problem solving. To address these challenges, we present PAPEL, a parent-AI collaborative system that grounds suggestions in the ongoing play scene and organizes support into four core modules: content generation, language adaptation, balance assessment, and extended response. In a counterbalanced within-subjects study with 16 parent-child dyads, PAPEL was associated with more integrated parent utterances that combined playful and instructional content, as well as more parent-child conversational turns, than the lightweight chatbot baseline used in our study.

cs.HC↗

arXiv2026

PACEE: Parent-Centered AI Scaffolding for Emotion Education in Early Childhood Conversations

Emotion education is critical for children aged 3 to 6. However, existing technologies largely focus on children's direct interaction with AI, overlooking the central role of parents in guiding early emotional development at home. To address this gap, we conducted co-design sessions with five kindergarten teachers and five parents to identify key parental challenges and opportunities for AI support in family emotion education. Based on these insights, we developed PACEE, an LLM-based assistant designed to support parents in guiding children's emotional development through conversations, rather than directly interacting with children. PACEE provides parent-centered AI scaffolding that supports parents in real-time conversation through personalized guidance, post-hoc reflection through trackable feedback, and understanding children's emotional states through modeling. We evaluated PACEE with 16 families. Results show that PACEE enhances parent-child engagement, fosters deeper emotional communication, and improves parents' expertise and overall experience in guiding their children. Our findings extend emotion coaching practices to the context of generative AI and offer design insights for building AI systems that support parent-centered family education.

cs.HC↗

arXiv2025

MOAT: Evaluating LMMs for Capability Integration and Instruction Grounding

Large multimodal models (LMMs) have demonstrated significant potential as generalists in vision-language (VL) tasks. However, adoption of LMMs in real-world tasks is hindered by their poor performance in tasks that require a combination of VL capabilities, as well as in tasks that involve the grounding of complex text or visual instructions. To thoroughly investigate this gap and its underlying causes, we propose MOAT, a diverse benchmark with 1005 complex real-world vision questions that are straightforward for humans but challenging for LMMs. Specifically, the tasks in MOAT require LMMs to engage in generalist problem solving by integrating VL capabilities such as reading text, counting, understanding spatial relations, grounding textual and visual instructions, etc. All these abilities fit into a taxonomy proposed by us that contains 9 VL capabilities, enabling MOAT to provide a fine-grained view of LMMs' strengths and weaknesses. Besides, MOAT is the first benchmark to explicitly evaluate LMMs' ability to ground complex text and visual instructions, which is essential for many real-world applications. We evaluated 17 proprietary and open source LMMs, finding that the best performing LMM (Gemini 2.5 Pro) achieved only 44% accuracy, far below what would be acceptable in real-world applications. To guide future model development, we analyze common trends in our results and discuss the underlying causes of poor performance, focusing on the impact of text-centric reasoning, which VL capabilities form bottlenecks in complex tasks, and the potential harmful effects of tiling. Code and data are available at https://cambrian-yzt.github.io/MOAT/.

cs.CL↗

arXiv2024

Perovskite topological exciton-polariton disclination laser at room temperature

Topologically nontrivial systems can be protected by band topology in momentum space, as seen in topological insulators and semimetals, or real-space topology, such as in lattice deformations known as topological disclinations (TDs). TDs, with inherent chiral symmetry, can support localized states pinned spectrally to the middle of the topological gap, preventing hybridization with bulk bands, and making them promising for topological lasers. Here, we experimentally realize a C4v symmetric TD laser based on perovskite exciton-polariton lattices at room temperature. Protected by the chiral and point group symmetries of the lattice, the TD state emerges in the middle of the gap and at the core of the perovskite lattice. Under a non-resonant pulsed excitation, coherent polariton lasing occurs precisely at the TD state with a low threshold of 9.5 uJ/cm2, as confirmed by momentum space and real space spectra measurements. This study not only introduces a class of symmetry-protected topological lasers, but also expands the landscape for exploring exciton-polariton light-matter interactions with novel topological structures.

physics.optics↗

arXiv2022

EvilModel 2.0: Bringing Neural Network Models into Malware Attacks

Security issues have gradually emerged with the continuous development of artificial intelligence (AI). Earlier work verified the possibility of converting neural network models into stegomalware, embedding malware into a model with limited impact on the model's performance. However, existing methods are not applicable in real-world attack scenarios and do not attract enough attention from the security community due to performance degradation and additional workload. Therefore, we propose an improved stegomalware EvilModel. By analyzing the composition of the neural network model, three new methods for embedding malware into the model are proposed: MSB reservation, fast substitution, and half substitution, which can embed malware that accounts for half of the model's volume without affecting the model's performance. We built 550 EvilModels using ten mainstream neural network models and 19 malware samples. The experiment shows that EvilModel achieved an embedding rate of 48.52\%. A quantitative algorithm is proposed to evaluate the existing embedding methods. We also design a trigger and propose a threat scenario for the targeted attack. The practicality and effectiveness of the proposed methods were demonstrated by experiments and analyses of the embedding capacity, performance impact, and detection evasion.

cs.CR↗