We present RegHead, a framework for constructing semantic blendshape sets for animatable non-humanoid head avatars. With a fixed expression vocabulary, semantic blendshapes provide a low-dimensional and interpretable animation interface and support cross-identity retargeting. Building such blendshape sets remains expensive because (i) expression-consistent supervision is scarce, (ii) generated 4D assets typically lack correspondence, and (iii) facial motion is highly localized. We propose (1) a large-scale dataset of non-humanoid identities paired with a shared expression vocabulary, obtained by expanding a small artist-rigged library via fine-tuned image editing; (2) a dense stochastic anchor motion representation tailored to localized facial deformations; and (3) a fast feed-forward registration model that converts unregistered expression meshes into a corresponded blendshape basis by predicting anchor-based deformations from the neutral shape. Experiments show that our approach produces higher-fidelity expression meshes than baselines, while running orders of magnitude faster than optimization. We further demonstrate real-time retargeting from human face tracking signals to non-humanoid characters, capturing both head pose and localized facial motions. Our project page is available at https://snap-research.github.io/RegHead/.

Guideline-consistent semantic segmentation requires more than category recognition, as real-world labeling policies demand fine-grained, task-specific decisions. Recent multi-agent refinement systems improve compliance with such textual guidelines by detecting and correcting errors. However, they are stateless: feedback from the critiquing agent is discarded, causing the same guideline-specific mistakes to be repeatedly rediscovered and corrected across the dataset at the cost of additional refinement. We introduce InsightSeg, an episodic memory mechanism that converts successful correction episodes into reusable, visually grounded insights. A meta-analyzer distills each qualifying episode into directive natural-language insights and anchors them to the local image regions that caused the error using patch-level visual concept vectors. On subsequent images, these concepts are matched against dense patch embeddings to retrieve relevant insights, which condition the segmenting agent before making its first prediction. This shifts the system from correcting recurring errors to preventing them, improving segmentation quality before any refinement occurs. Across Waymo and Cityscapes, InsightSeg improves both first-pass and final guideline-consistent segmentation performance while requiring fewer refinement steps, demonstrating that multi-agent refinement can become more accurate and efficient by drawing on past correction experience.

Semantic segmentation in real-world applications often requires not only accurate masks but also strict adherence to textual labeling guidelines. These guidelines are typically complex and long, and both human and automated labeling often fail to follow them faithfully. Traditional approaches depend on expensive task-specific retraining that must be repeated as the guidelines evolve. Although recent open-vocabulary segmentation methods excel with simple prompts, they often fail when confronted with sets of paragraph-length guidelines that specify intricate segmentation rules. To address this, we introduce a multi-agent, training-free framework that coordinates general-purpose vision-language models within an iterative Worker-Supervisor refinement architecture. The Worker performs the segmentation, the Supervisor critiques it against the retrieved guidelines, and a lightweight reinforcement learning stop policy decides when to terminate the loop, ensuring guideline-consistent masks while balancing resource use. Evaluated on the Waymo and ReasonSeg datasets, our method notably outperforms state-of-the-art baselines, demonstrating strong generalization and instruction adherence.

Text-to-image(T2I) models like Stable Diffusion and DALL-E have made generative AI widely accessible, yet recent studies reveal that these systems often replicate societal biases, particularly in how they depict demographic groups across professions. Prompts such as 'doctor' or 'CEO' frequently yield lighter-skinned outputs, while lower-status roles like 'janitor' show more diversity, reinforcing stereotypes. Existing mitigation methods typically require retraining or curated datasets, making them inaccessible to most users. We propose a lightweight, inference-time framework that mitigates representational bias through prompt-level intervention without modifying the underlying model. Instead of assuming a single definition of fairness, our approach allows users to select among multiple fairness specifications-ranging from simple choices such as a uniform distribution to more complex definitions informed by a large language model(LLM) that cites sources and provides confidence estimates. These distributions guide the construction of demographic specific prompt variants in the corresponding proportions, and we evaluate alignment by auditing adherence to the declared target and measuring the resulting skin tone distribution rather than assuming uniformity as 'fairness'. Across 36 prompts spanning 30 occupations and 6 non-occupational contexts, our method shifts observed skin-tone outcomes in directions consistent with the declared target, and reduces deviation from targets when the target is defined directly in skin-tone space(fallback). This work demonstrates how fairness interventions can be made transparent, controllable, and usable at inference time, directly empowering users of generative AI.

As the capabilities of artificial intelligence (AI) continue to expand rapidly, Human-AI (HAI) Collaboration, combining human intellect and AI systems, has become pivotal for advancing problem-solving and decision-making processes. The advent of Large Foundation Models (LFMs) has greatly expanded its potential, offering unprecedented capabilities by leveraging vast amounts of data to understand and predict complex patterns. At the same time, realizing this potential responsibly requires addressing persistent challenges related to safety, fairness, and control. This paper reviews the crucial integration of LFMs with HAI, highlighting both opportunities and risks. We structure our analysis around: human-guided model development, collaborative design principles, ethical and governance frameworks, and applications in high-stakes domains. Our review shows that successful HAI systems are not the automatic result of stronger models but the product of careful, human-centered design. By identifying key open challenges, this survey aims to give insight into current and future research that turns the raw power of LFMs into partnerships that are reliable, trustworthy, and beneficial to society.

We present a novel generative model that synthesizes photorealistic, biophysically plausible faces by capturing the intricate relationships between facial geometry and biophysical attributes. Our approach models facial appearance in a biophysically grounded manner, allowing for the editing of both high-level attributes such as age and gender, as well as low-level biophysical properties such as melanin level and blood content. This enables continuous modeling of physical skin properties that correlate changes in skin properties with shape changes. We showcase the capabilities of our framework beyond its role as a generative model through two practical applications: editing the texture maps of 3D faces that have already been captured, and serving as a strong prior for face reconstruction when combined with differentiable rendering. Our model allows for the creation of physically-based relightable, editable faces with consistent topology and uv layout that can be integrated into traditional computer graphics pipelines.

Avatar creation from human images presents challenges for direct neural approaches, which suffer from inconsistent predictions and poor interpretability due to the large parameter space with hundreds of ambiguous options. We propose a neurosymbolic tag-based annotation method that combines neural perceptual learning with symbolic semantic reasoning. Instead of directly predicting avatar parameters, our approach uses a neural network to predict semantic tags (hair length, curliness, direction) as an intermediate symbolic representation, then applies symbolic search algorithms to match optimal avatar assets. This neurosymbolic design produces higher annotator agreements (96.7% vs 31.0% for direct annotation), enables more consistent model predictions, and provides interpretable avatar selection with ranked alternatives. The tag-based system generalizes easily across rendering systems, requiring only new asset annotation while reusing human image tags. Experimental results demonstrate superior convergence, consistency, and visual quality compared to direct prediction methods, showing how neurosymbolic approaches can improve trustworthiness and interpretability in creative AI applications.

This article presents RipScout, a system for realtime rip current detection and data collection using drones equipped with machine learning (ML). No internet connection is required. RipScout achieves realtime performance by using lightweight ML models that fit the constraints of limited mobile computing resources with the drone controller. We compared different ML models trained to detect either one or two types of rip currents. The best model was then selected for RipScout. The system was evaluated with three ML models, with the EfficientDet D2 model achieving the highest accuracy of 93.1% for multiclass detection while maintaining realtime processing at an average speed of 17 frames per second. When a rip is detected along a flight path, the drone hovers in place and collects a video clip of a predefined length, followed by circling around the detected rip using prespecified radii and heights to collect video samples from different vantage points and elevations. An important benefit of RipScout is that the collection of rip current data can be performed by drone operators who are not familiar with rip currents. We conducted field tests and found that the proposed system allows data to be collected four times faster than without it while improving accuracy. As a by-product of the field experiments, we also provide a new rip current dataset. Such a multiviewpoint dataset can be used to improve rip current detection, especially from lower elevations and different orientations.

We present SplatFace, a novel Gaussian splatting framework designed for 3D human face reconstruction without reliance on accurate pre-determined geometry. Our method is designed to simultaneously deliver both high-quality novel view rendering and accurate 3D mesh reconstructions. We incorporate a generic 3D Morphable Model (3DMM) to provide a surface geometric structure, making it possible to reconstruct faces with a limited set of input images. We introduce a joint optimization strategy that refines both the Gaussians and the morphable surface through a synergistic non-rigid alignment process. A novel distance metric, splat-to-surface, is proposed to improve alignment by considering both the Gaussian position and covariance. The surface information is also utilized to incorporate a world-space densification process, resulting in superior reconstruction quality. Our experimental analysis demonstrates that the proposed method is competitive with both other Gaussian splatting techniques in novel view synthesis and other 3D reconstruction methods in producing 3D face meshes with high geometric precision.

Large Language Model (LLM) agents are increasingly used in many applications, raising concerns about their safety. While previous work has shown that LLMs can deceive in controlled tasks, less is known about their ability to deceive using natural language in social contexts. In this paper, we study deception in the Social Deduction Game (SDG) Mafia, where success is dependent on deceiving others through conversation. Unlike previous SDG studies, we use an asynchronous multi-agent framework which better simulates realistic social contexts. We simulate 35 Mafia games with GPT-4o LLM agents. We then create a Mafia Detector using GPT-4-Turbo to analyze game transcripts without player role information to predict the mafia players. We use prediction accuracy as a surrogate marker for deception quality. We compare this prediction accuracy to that of 28 human games and a random baseline. Results show that the Mafia Detector's mafia prediction accuracy is lower on LLM games than on human games. The result is consistent regardless of the game days and the number of mafias detected. This indicates that LLMs blend in better and thus deceive more effectively. We also release a dataset of LLM Mafia transcripts to support future research. Our findings underscore both the sophistication and risks of LLM deception in social contexts.

Medical image segmentation is crucial for diagnosis and treatment planning. Traditional CNN-based models, like U-Net, have shown promising results but struggle to capture long-range dependencies and global context. To address these limitations, we propose a transformer-based architecture that jointly applies Channel Attention and Pyramid Attention mechanisms to improve multi-scale feature extraction and enhance segmentation performance for medical images. Increasing model complexity requires more training data, and we further improve model generalization with CutMix data augmentation. Our approach is evaluated on the Synapse multi-organ segmentation dataset, achieving a 6.9 improvement in Mean Dice score and a 39.9 (HD95) over an implementation without our enhancements. Our proposed model demonstrates improved segmentation accuracy for complex anatomical structures, outperforming existing state-of-the-art methods.

Background:Population health surveys are an important tool to effectively allocate limited resources in low-resource communities. In such an environment, surveys are often done by the local population with pen and paper. Data thus collected are difficult to tabulate and analyze. Objective:The objective of this study was to evaluate the viability and efficiency of mobile forms as an alternative to paper-based surveys in a specific low-resource setting. Methods:We conducted pilot interviews with 53 local surveyors in the Philippines to assess their initial attitudes toward mobile forms. We then built software that can generate mobile forms that are easy to use, capable of working offline, and able to track key metrics such as time to complete questions. Our mobile form was field-tested in 3 locations in the Philippines with 33 surveyors collecting health survey responses from 266 participants. Results:In the pilot phase, we found that 32 out of 53 (60%) of the local surveyors preferred mobile forms over paper. After field-testing, the number of surveyors preferring mobile forms increased to 25 out of 33 (76%) after just using the form a few times. The mobile forms overall demonstrated enhanced efficiency in data collection and usability over paper surveys. Conclusions:Our findings indicate that mobile forms are a viable method to conduct large-scale population health surveys in this low-resource environment.

We present T2Bs, a framework for generating high-quality, animatable character head morphable models from text by combining static text-to-3D generation with video diffusion. Text-to-3D models produce detailed static geometry but lack motion synthesis, while video diffusion models generate motion with temporal and multi-view geometric inconsistencies. T2Bs bridges this gap by leveraging deformable 3D Gaussian splatting to align static 3D assets with video outputs. By constraining motion with static geometry and employing a view-dependent deformation MLP, T2Bs (i) outperforms existing 4D generation methods in accuracy and expressiveness while reducing video artifacts and view inconsistencies, and (ii) reconstructs smooth, coherent, fully registered 3D geometries designed to scale for building morphable models with diverse, realistic facial motions. This enables synthesizing expressive, animatable character heads that surpass current 4D generation techniques.

There are 50 billion pieces of litter in the U.S. alone. Grass fields contribute to this problem because picnickers tend to leave trash on the field. We propose building a robot that can autonomously navigate, identify, and pick up trash in parks. To autonomously navigate the park, we used a Spanning Tree Coverage (STC) algorithm to generate a coverage path the robot could follow. To navigate this path, we successfully used Real-Time Kinematic (RTK) GPS, which provides a centimeter-level reading every second. For computer vision, we utilized the ResNet50 Convolutional Neural Network (CNN), which detects trash with 94.52

Open source software development, particularly within institutions such as universities and research laboratories, is often decentralized and difficult to track. Although academic teams produce many impactful scientific tools, their projects do not always follow consistent open source practices, such as clear licensing, documentation, or community engagement. As a result, these efforts often go unrecognized due to limited visibility and institutional awareness, and the software itself can be difficult to sustain over time. This paper presents an end-to-end framework for systematically discovering and analyzing open source projects across distributed academic systems. Using ten universities as a case study, we build a pipeline that collects data via GitHub's REST API, extracts metadata, and predicts both institutional affiliation and project type (e.g., development tools, educational materials, websites, documentation). Applied across the ten campuses, our method identifies over 200,000 repositories and collects information on their activity and open source practices, enabling a deeper understanding of institutional open source contributions. Beyond discovery, our framework enables actionable insights into institutional open source practices, revealing patterns such as missing licenses or limited community engagement. These findings can guide targeted support, policy development, and strategies to strengthen open source contributions across academic institutions.

Rip currents present a significant safety risk to beach tourists and coastal communities, resulting in hundreds of annual drownings all over the world. A key contributing factor to this danger is the lack of awareness among beachgoers about recognizing and avoiding these rip currents. In response to this issue, we introduce RipFinder, a mobile app equipped with machine learning (ML) models trained to detect two types of rip currents. Users can leverage the app’s computer vision capabilities to use their phone’s camera to identify these hazardous rip currents in real time. The amorphous and ephemeral nature of rip currents makes it challenging to detect them with high accuracy using object detection models. To address this, we propose a client-server ML model-based computer vision system designed specifically to improve rip current detection accuracy. This novel approach enables the app to function with or without internet connectivity, proving particularly beneficial in regions without lifeguards or internet access. Additionally, the app serves as an educational resource, offering in-app information about rip currents. It also promotes citizen science involvement by encouraging users to contribute valuable information on detected rip currents. This paper presents the app’s overall design and discusses the challenges inherent to the rip current detection system.

Vision-language models (VLMs) have shown strong performance on text-to-image retrieval benchmarks. However, bridging this success to real-world applications remains a challenge. In practice, human search behavior is rarely a one-shot action. Instead, it is often a multi-round process guided by clues in mind. That is, a mental image ranging from vague recollections to vivid mental representations of the target image. Motivated by this gap, we study the task of Mental Image Retrieval (MIR), which targets the realistic yet underexplored setting where users refine their search for a mentally envisioned image through multi-round interactions with an image search engine. Central to successful interactive retrieval is the capability of machines to provide users with clear, actionable feedback; however, existing methods rely on indirect or abstract verbal feedback, which can be ambiguous, misleading, or ineffective for users to refine the query. To overcome this, we propose GenIR, a generative multi-round retrieval paradigm leveraging diffusion-based image generation to explicitly reify the AI system's understanding at each round. These synthetic visual representations provide clear, interpretable feedback, enabling users to refine their queries intuitively and effectively. We further introduce a fully automated pipeline to generate a high-quality multi-round MIR dataset. Experimental results demonstrate that GenIR significantly outperforms existing interactive methods in the MIR scenario. This work establishes a new task with a dataset and an effective generative retrieval method, providing a foundation for future research in this direction

Vehicle object detection benefits from both LiDAR and camera data, with LiDAR offering superior performance in many scenarios. Fusion of these modalities further enhances accuracy, but existing methods often introduce complexity or dataset-specific dependencies. In our study, we propose a model-adaptive late-fusion method, VaLID, which validates whether each predicted bounding box is acceptable or not. Our method verifies the higher-performing, yet overly optimistic LiDAR model detections using camera detections that are obtained from either specially trained, general, or open-vocabulary models. VaLID uses a lightweight neural verification network trained with a high recall bias to reduce the false predictions made by the LiDAR detector, while still preserving the true ones. Evaluating with multiple combinations of LiDAR and camera detectors on the KITTI dataset, we reduce false positives by an average of 63.9%, thus outperforming the individual detectors on 3D average precision (3DAP). Our approach is model-adaptive and demonstrates state-of-the-art competitive performance even when using generic camera detectors that were not trained specifically for this dataset.

Artificial intelligence (AI) models for computer vision trained with supervised machine learning are assumed to solve classification tasks by imitating human behavior learned from training labels. Most efforts in recent vision research focus on measuring the model task performance using standardized benchmarks such as accuracy. However, limited work has sought to understand the perceptual difference between humans and machines. To fill this gap, this study first analyzes the statistical distributions of mistakes from the two sources and then explores how task difficulty level affects these distributions. We find that even when AI learns an excellent model from the training data, one that outperforms humans in overall accuracy, these AI models have significant and consistent differences from human perception. We demonstrate the importance of studying these differences with a simple human-AI teaming algorithm that outperforms humans alone, AI alone, or AI-AI teaming.