G3Ego: Gaze-Guided Graphs for Egocentric Action Understanding

Vision-Language Models for Egocentric Video: From Hand-Object Interaction to Embodied AI

This survey presents a critical review of VLMs for egocentric video understanding, tracing the progression from conventional recognition architectures to multimodal foundation models and embodied systems, and examines how first-person perception and multimodal foundation models support wearable assistance, robot skill learning, human-to-robot transfer, and embodied decision making.

SAGE: Synchronized Action-Gaze Recognition and Anticipation for Human Behavior Understanding

Human object interaction (HOI), gaze pattern, and their anticipation are intricately linked, providing valuable insights into cognitive processes, intentions, and behavior. However, most existing models handle gaze and actions separately, missing both their interdependence and the advantages of a unified solution. This paper presents a novel unified framework, SAGE (Synchronized Action-GazE), which integrates simultaneous recognition and anticipation of both HOI and human gaze into a single unified end-to-end trainable model. Our approach leverages a transformer-based architecture and incorporates gaze data into spatiotemporal attention mechanisms to simultaneously predict current and future human actions and gaze behavior. We explore this bidirectional relationship between gaze and actions under different scenarios, whether requiring a close-up, detailed view (egocentric) or a wider, more contextual view (exocentric), making our framework versatile for various applications. Additionally, due to lack of datasets for comprehensive analysis of both HOI and gaze in exocentric videos, we establish a new benchmark Exo-Cook to facilitate further research in this domain. Experiments on three benchmark datasets: VidHOI, EGTEA Gaze+, and Exo-Cook show that jointly modeling gaze and actions across current and future frames achieves consistently strong results, often surpassing specialized state-of-the-art models tailored to individual tasks. By unifying actions and attention in a comprehensive way, our work lays the groundwork for more intuitive human-machine interaction.

Token-Based Affordance Grounding with Large Vision-Language Models

Affordance grounding aims to localize image regions that support a specific action, serving as a core capability for physical intelligence and embodied perception. Previous studies have primarily relied on weakly supervised learning with action labels from exocentric images. However, these methods often struggle with visually ambiguous exocentric images containing co-occurring actions; moreover, they fail to distinguish semantically similar actions because existing methods typically rely on brief action phrases that lack rich semantic details for action-specific localization. Although large vision-language models (LVLMs) encode rich action semantics and their action-conditioned textual outputs implicitly contain spatial cues, they do not directly provide action-specific spatial localization. To address these problems, we propose TokAG, a zero-shot affordance grounding framework that exploits the token-level semantic-spatial signals in LVLMs to localize action-relevant regions without external supervision. We observe that attention maps associated with different LVLM output tokens vary significantly, with many attending to irrelevant regions such as the background. Thus, we introduce a spatial-aware token-selection mechanism to systematically evaluate each output token and select the one whose attention maps exhibit dominant activation over the target object, instead of relying on arbitrary attention maps. By extracting these object-focused attention maps, we transform the LVLM's implicit semantic signals into zero-shot affordance heatmaps. Our zero-shot framework consistently outperforms prior weakly supervised approaches across multiple benchmarks, improving NSS by 10.7% on the unseen split of AGD20K and by 29.7% on HICO-IIF. The code and models will be made publicly available.

LogFA: Feature-Space Data Augmentation in Long Egocentric Videos Using Foundation Models

GazeHRNet: Head-Centric Spatial Encoding and Gaze-Aware Feature Interaction for Gaze Target Detection

Gaze target detection requires understanding where a person is looking by jointly reasoning about the gazer and the surrounding scene. While recent methods have benefited from powerful pretrained visual backbones, they often treat gaze prediction as a generic localization problem and overlook a key property of the task: the target should be interpreted in relation to the person’s head. This limits their ability to model direction, distance, and head-scene dependencies in a unified manner. We propose GazeHRNet, a head-centric reasoning framework for RGB-based gaze target detection. Instead of relying on absolute image coordinates or auxiliary geometric inputs, GazeHRNet represents the scene from the gazer’s perspective through Head-Centric Polar Encoding and organizes visual features by their spatial relevance to the head via Head-Aware Attention Routing. It further combines coarse spatial reasoning with fine-grained anisotropic heatmap prediction, enabling reliable target localization under cluttered scenes and varying head positions. Experiments on GazeFollow and VideoAttentionTarget show that GazeHRNet achieves 0.952 and 0.929 AUC with L2 distances of 0.102 and 0.103, respectively, using only RGB input and 3 M trainable parameters. Cross-dataset evaluation further demonstrates improved robustness and generalization across different scenes and subject distributions.

Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?

Hand-object interaction (HOI) recognition requires capturing both hand manipulations and object transformations. However, existing video-language models often fall into shortcuts by relying on spurious correlations among hands, objects, or environmental context, rather than reasoning from the appearance and dynamics of hands and objects themselves. To address this limitation, we propose a new learning paradigm that combines (i) hand-object masked training, which enables robust reasoning from partial hand or object observations, and (ii) an HOI-dynamics-aware decoder that explicitly learns hand- and object-centric embeddings through auxiliary predictions of their locations and semantics, enhancing sensitivity to both cues. To systematically evaluate such cue-specific reasoning, we introduce Cue-Isolated HOI (CI-HOI), a new evaluation that assesses models'ability to predict actions from hand- and object-related cues independently. To enable CI-HOI, we curate the DEHOI testbed, which separates hand- and object-related observations for disentangled HOI evaluation through inpainting. Using DEHOI, we demonstrate both quantitatively and qualitatively that our training strategy exploits hand- and object-centric information more effectively than existing models. Our approach improves over existing models on DEHOI, standard action recognition, object state recognition, and even robot manipulation action recognition, leading to more robust HOI understanding.