Humans excel at spatial-temporal reasoning, effortlessly interpreting dynamic visual events from an egocentric viewpoint. However, whether multimodal large language models (MLLMs) can similarly understand the 4D world remains uncertain. This paper explores multimodal spatial-temporal reasoning from an egocentric perspective, aiming to equip MLLMs with human-like reasoning capabilities. To support this objective, we introduce Ego-ST Bench, a novel benchmark containing over 5,000 question-answer pairs across four categories, systematically evaluating spatial, temporal, and integrated spatial-temporal reasoning. Additionally, we propose ST-R1 training paradigm, a video-based reasoning model that incorporates reverse thinking into its reinforcement learning process, significantly enhancing performance. We combine long-chain-of-thought (long-CoT) supervised fine-tuning with Group Relative Policy Optimization (GRPO) reinforcement learning, achieving notable improvements with limited high-quality data. Ego-ST Bench and ST-R1 provide valuable insights and resources for advancing video-based spatial-temporal reasoning research.

With the rapid advancement of AI systems, their abilities to store, retrieve, and utilize information over the long term - referred to as long-term memory - have become increasingly significant. These capabilities are crucial for enhancing the performance of AI systems across a wide range of tasks. However, there is currently no comprehensive survey that systematically investigates AI's long-term memory capabilities, formulates a theoretical framework, and inspires the development of next-generation AI long-term memory systems. This paper begins by introducing the mechanisms of human long-term memory, then explores AI long-term memory mechanisms, establishing a mapping between the two. Based on the mapping relationships identified, we extend the current cognitive architectures and propose the Cognitive Architecture of Self-Adaptive Long-term Memory (SALM). SALM provides a theoretical framework for the practice of AI long-term memory and holds potential for guiding the creation of next-generation long-term memory driven AI systems. Finally, we delve into the future directions and application prospects of AI long-term memory.

Long-form video understanding with Large Vision Language Models is challenged by the need to analyze temporally dispersed yet spatially concentrated key moments within limited context windows. In this work, we introduce VideoMindPalace, a new framework inspired by the "Mind Palace", which organizes critical video moments into a topologically structured semantic graph. VideoMindPalace organizes key information through (i) hand-object tracking and interaction, (ii) clustered activity zones representing specific areas of recurring activities, and (iii) environment layout mapping, allowing natural language parsing by LLMs to provide grounded insights on spatio-temporal and 3D context. In addition, we propose the Video Mind-Palace Benchmark (VMB), to assess human-like reasoning, including spatial localization, temporal reasoning, and layout-aware sequential understanding. Evaluated on VMB and established video QA datasets, including EgoSchema, NExT-QA, IntentQA, and the Active Memories Benchmark, VideoMindPalace demonstrates notable gains in spatiotemporal coherence and human-aligned reasoning, advancing long-form video analysis capabilities in VLMs. Project website: https://oodbag.github.io/VMP/.

Text-guided image manipulation has experienced notable advancement in recent years. In order to mitigate linguistic ambiguity, few-shot learning with visual examples has been applied for instructions that are underrepresented in the training set, or difficult to describe purely in language. However, learning from visual prompts requires strong reasoning capability, which diffusion models are struggling with. To address this issue, we introduce a novel multi-modal autoregressive model, dubbed InstaManip, that can instantly learn a new image manipulation operation from textual and visual guidance via in-context learning, and apply it to new query images. Specifically, we propose an innovative group self-attention mechanism to break down the in-context learning process into two separate stages – learning and applying, which simplifies the complex problem into two easier tasks. We also introduce a relation regularization method to further disentangle image transformation features from irrelevant contents in exemplar images. Extensive experiments suggest that our method surpasses previous few-shot image manipulation models by a notable margin (≥19% in human evaluation). We also find our model can be further boosted by increasing the number or diversity of exemplar images. Please check out our project page (https://bolinlai.github.io/projects/InstaManip/).

Prevailing Multimodal Large Language Models (MLLMs) encode the input image(s) as vision tokens and feed them into the language backbone, similar to how Large Language Models (LLMs) process the text tokens. However, the number of vision tokens increases quadratically as the image resolutions, leading to huge computational costs. In this paper, we consider improving MLLM’s efficiency from two scenarios, (I) Reducing computational cost without degrading the performance. (II) Improving the performance with given budgets. We start with our main finding that the ranking of each vision token sorted by attention scores is similar in each layer except the first layer. Based on it, we assume that the number of essential top vision tokens does not increase along layers. Accordingly, for Scenario I, we propose a greedy search algorithm (G-Search) to find the least number of vision tokens to keep at each layer from the shallow to the deep. Interestingly, G-Search is able to reach the optimal reduction strategy based on our assumption. For Scenario II, based on the reduction strategy from G-Search, we design a parametric sigmoid function (P-Sigmoid) to guide the reduction at each layer of the MLLM, whose parameters are optimized by Bayesian Optimization. Extensive experiments demonstrate that our approach can significantly accelerate those popular MLLMs, e.g. LLaVA, and InternVL2 models, by more than 2⇥ without performance drops. Our approach also far outperforms other token reduction methods when budgets are limited, achieving a better trade-off between efficiency and effectiveness.

We present Ego-Exo4D, a diverse, large-scale multimodal multiview video dataset and benchmark challenge. Ego-Exo4D centers around simultaneously-captured egocentric and exocentric video of skilled human activities (e.g., sports, music, dance, bike repair). 740 participants from 13 cities worldwide performed these activities in 123 different natural scene contexts, yielding long-form captures from 1 to 42 minutes each and 1,286 hours of video combined. The multimodal nature of the dataset is unprecedented: the video is accompanied by multichannel audio, eye gaze, 3D point clouds, camera poses, IMU, and multiple paired language descriptions—including a novel “expert commentary” done by coaches and teachers and tailored to the skilled-activity domain. To push the frontier of first-person video understanding of skilled human activity, we also present a suite of benchmark tasks and their annotations, including fine-grained activity understanding, proficiency estimation, cross-view translation, and 3D hand/body pose. All resources are open sourced to fuel new research in the community. https://ego-exo4d-data.org/

We introduce Ego4D, a massive-scale egocentric video dataset and benchmark suite. It offers 3,670 hours of daily-life activity video spanning hundreds of scenarios (household, outdoor, workplace, leisure, etc.) captured by 931 unique camera wearers from 74 worldwide locations and 9 different countries. The approach to collection is designed to uphold rigorous privacy and ethics standards, with consenting participants and robust de-identification procedures where relevant. Ego4D dramatically expands the volume of diverse egocentric video footage publicly available to the research community. Portions of the video are accompanied by audio, 3D meshes of the environment, eye gaze, stereo, and/or synchronized videos from multiple egocentric cameras at the same event. Furthermore, we present a host of new benchmark challenges centered around understanding the first-person visual experience in the past (querying an episodic memory), present (analyzing hand-object manipulation, audio-visual conversation, and social interactions), and future (forecasting activities). By publicly sharing this massive annotated dataset and benchmark suite, we aim to push the frontier of first-person perception.

Traditionally, speech quality evaluation relies on subjective assessments or intrusive methods that require reference signals or additional equipment. However, over recent years, non-intrusive speech quality assessment has emerged as a promising alternative, capturing much attention from researchers and industry professionals. This article presents a deep learning-based method that exploits large-scale intrusive simulated data to improve the accuracy and generalization of non-intrusive methods. The major contributions of this article are as follows. First, it presents a data simulation method, which generates degraded speech signals and labels their speech quality with the perceptual objective listening quality assessment (POLQA). The generated data is proven to be useful for pretraining the deep learning models. Second, it proposes to apply an adversarial speaker classifier to reduce the impact of speaker-dependent information on speech quality evaluation. Third, an autoencoder-based deep learning scheme is proposed following the principle of representation learning and adversarial training (AT) methods, which is able to transfer the knowledge learned from a large amount of simulated speech data labeled by POLQA. With the help of discriminative representations extracted from the autoencoder, the prediction model can be trained well on a relatively small amount of speech data labeled through subjective listening tests. Fourth, an end-to-end speech quality evaluation neural network is developed, which takes magnitude and phase spectral features as its inputs. This phase-aware model is more accurate than the model using only the magnitude spectral features. A large number of experiments are carried out with three datasets: one simulated with labels obtained using POLQA and two recorded with labels obtained using subjective listening tests. The results show that the presented phase-aware method improves the performance of the baseline model and the proposed model with latent representations extracted from the adversarial autoencoder (AAE) outperforms the state-of-the-art objective quality assessment methods, reducing the root mean square error (RMSE) by 10.5% and 12.2% on the Beijing Institute of Technology (BIT) dataset and Tencent Corpus, respectively. The code and supplementary materials are available at https://github.com/liushenme/AAE-SQA.

The systematic evaluation and understanding of computer vision models under varying conditions require large amounts of data with comprehensive and customized labels, which real-world vision datasets rarely satisfy. While current synthetic data generators offer a promising alternative, particularly for embodied AI tasks, they often fall short for computer vision tasks due to low asset and rendering quality, limited diversity, and unrealistic physical properties. We introduce the BEHAVIOR Vision Suite (BVS), a set of tools and assets to generate fully customized synthetic data for systematic evaluation of computer vision models, based on the newly developed embodied AI benchmark, BEHAVIOR-1 K. BVS supports a large number of adjustable parameters at the scene level (e.g., lighting, object placement), the object level (e.g., joint configuration, attributes such as “filled” and “folded”), and the camera level (e.g., field of view, focal length). Researchers can arbitrarily vary these parameters during data generation to perform controlled experiments. We showcase three example application scenarios: systematically evaluating the robustness of models across different continuous axes of domain shift, evaluating scene understanding models on the same set of images, and training and evaluating simulation-to-real transfer for a novel vision task: unary and binary state prediction. Project website: https://behavior-vision-suite.github.io/

We present Ego-Exo4D, a diverse, large-scale multi-modal multiview video dataset and benchmark challenge. Ego-Exo4D centers around simultaneously-captured ego-centric and exocentric video of skilled human activities (e.g., sports, music, dance, bike repair). 740 participants from 13 cities worldwide performed these activities in 123 different natural scene contexts, yielding long-form captures from 1 to 42 minutes each and 1,286 hours of video combined. The multimodal nature of the dataset is un-precedented: the video is accompanied by multichannel audio, eye gaze, 3D point clouds, camera poses, IMU, and multiple paired language descriptions-including a novel “expert commentary” done by coaches and teachers and tailored to the skilled-activity domain. To push the frontier of first-person video understanding of skilled human activity, we also present a suite of benchmark tasks and their annotations, including fine-grained activity understanding, proficiency estimation, cross-view translation, and 3D hand/body pose. All resources are open sourced to fuel new research in the community.

In recent years, the thriving development of research related to egocentric videos has provided a unique perspective for the study of conversational interactions, where both visual and audio signals play a crucial role. While most prior work focus on learning about behaviors that directly involve the camera wearer, we introduce the Ego-Exocentric Conversational Graph Prediction problem, marking the first attempt to infer exocentric conversational interactions from egocentric videos. We propose a unified multi-modal framework-Audio- Visual Conversational Attention (AV-CONV), for the joint prediction of conversation behaviors-speaking and listening-for both the camera wearer as well as all other social partners present in the egocentric video. Specifically, we adopt the self-attention mechanism to model the representations across-time, across-subjects, and across-modalities. To validate our method, we conduct experiments on a challenging egocentric video dataset that includes multi-speaker and multi-conversation scenarios. Our results demonstrate the superior performance of our method compared to a series of baselines. We also present detailed ablation studies to assess the contribution of each component in our model. Check our Project Page.

With the growing significance of non-intrusive speech quality assessment in speech systems, existing methods predominantly rely on neural networks to extract low-order features. Typically, these features undergo a low-dimensional linear transformation, yielding the network’s output. However, the intercorrelation between feature points is often overlooked. In this paper, we explore the concept of kernel method, which maps features into high dimensional space through dot product, in order to enhance the extraction of relationships among all feature points. Considering the unique advantages of tensors in complex data representation, we extend the utilization of tensor network and propose a novel framework that incorporates a matrix product state (MPS) layer to predict mean opinion score (MOS). By integrating the MPS layer, our model can transform low-order features into higher-order representations, facilitating linear transformation in a high dimensional space without increasing the number of parameters. Furthermore, we propose a loss function that concurrently assesses regression and classification biases, along with correlation with real MOS labels. Experimental results demonstrate that our proposed model consistently outperforms the baseline system across all evaluation metrics and surpasses state-of-the-art models on the test set.

Binaural audio delivers an immersive spatial auditory experience to human listeners, but most existing videos lack binaural audio due to the expertise required for recording environments. Recent studies have been dedicated to converting monaural audio into binaural ones conditioned on the visual inputs. In this paper, we propose a novel audio-visual spatialization network with two added audio decoders, which rely on carefully designed visual features to generate audio outputs for the left and right channels, respectively. In addition, we propose an audio-visual matching loss to further explore the correlation between binaural audio and the scene visual input. Experiment results show that the proposed method outperforms several state-of-the-art binaural audio generation methods on two benchmark datasets FAIR-Play and MUSIC-Stereo. Qualitative results are also presented to demonstrate the effectiveness of the proposed method.

As one of the crucial elements in human-robot interaction, responsive listening head generation has attracted considerable attention from researchers. It aims to generate a listening head video based on speaker's audio and video as well as a reference listener image. However, existing methods exhibit two limitations: 1) the generation capability of their models is limited, resulting in generated videos that are far from real ones, and 2) they mostly employ autoregressive generative models, unable to mitigate the risk of error accumulation. To tackle these issues, we propose Listenformer that leverages the powerful temporal modeling capability of transformers for generation. It can perform non-autoregressive prediction with the proposed two-stage training method, simultaneously achieving temporal continuity and overall consistency in the outputs. To fully utilize the information from the speaker inputs, we designed an audio-motion attention fusion module, which improves the correlation of audio and motion features for accurate response. Additionally, a novel decoding method called sliding window with a large shift is proposed for Listenformer, demonstrating both excellent computational efficiency and effectiveness. Extensive experiments show that Listenformer outperforms the existing state-of-the-art methods on ViCo and L2L datasets. And a perceptual user study demonstrates the comprehensive performance of our method in generating diversity, identity preserving, speaker-listener synchronization, and attitude matching. Our code is available at https://liushenme.github.io/ListenFormer.github.io/.

Predicting human's gaze from egocentric videos serves as a critical role for human intention understanding in daily activities. In this paper, we present the first transformer-based model to address the challenging problem of egocentric gaze estimation. We observe that the connection between the global scene context and local visual information is vital for localizing the gaze fixation from egocentric video frames. To this end, we design the transformer encoder to embed the global context as one additional visual token and further propose a novel global-local correlation module to explicitly model the correlation of the global token and each local token. We validate our model on two egocentric video datasets - EGTEA Gaze + and Ego4D. Our detailed ablation studies demonstrate the benefits of our method. In addition, our approach exceeds the previous state-of-the-art model by a large margin. We also apply our model to a novel gaze saccade/fixation prediction task and the traditional action recognition problem. The consistent gains suggest the strong generalization capability of our model. We also provide additional visualizations to support our claim that global-local correlation serves a key representation for predicting gaze fixation from egocentric videos. More details can be found in our website (https://bolinlai.github.io/GLC-EgoGazeEst).

Audio-visual deepfake detection is the process of identifying and detecting deepfakes that have been generated using both audio and visual content with AI algorithms. Most existing methods primarily focus on the overall authenticity while neglecting the position of forgeries in time. This can be particularly problematic, as even a small alteration in a clip can significantly impact its meaning. Such brand new attacks are dangerous and how to tackle such attacks remains an open question. In this paper, we present a novel neural network-based model to tackle the temporal forgery detection (TFD) problem. It consists of new audio and visual encoders with cross-modal attention for embedding extraction, and an embedding-level fusion mechanism with self-attention for forgery localization. Besides, a multi-dimensional contrastive loss is proposed which helps the model not only to capture audio-visual inconsistency for deepfake detection but also to exploit temporal inconsistency by coherently constraining the extracted embeddings. Extensive experiments on the LAV-DF dataset show that the presented method outperforms several state-of-the-art temporal forgery localization methods by up to 23.4% on [email protected] and 13.8% on AR@100. In addition, we also show the effectiveness of the proposed model on deepfake detection.

We address the task of jointly determining what a person is doing and where they are looking based on the analysis of video captured by a headworn camera. To facilitate our research, we first introduce the EGTEA Gaze+ dataset. Our dataset comes with videos, gaze tracking data, hand masks and action annotations, thereby providing the most comprehensive benchmark for First Person Vision (FPV). Moving beyond the dataset, we propose a novel deep model for joint gaze estimation and action recognition in FPV. Our method describes the participant's gaze as a probabilistic variable and models its distribution using stochastic units in a deep network. We further sample from these stochastic units, generating an attention map to guide the aggregation of visual features for action recognition. Our method is evaluated on our EGTEA Gaze+ dataset and achieves a performance level that exceeds the state-of-the-art by a significant margin. More importantly, we demonstrate that our model can be applied to larger scale FPV dataset-EPIC-Kitchens even without using gaze, offering new state-of-the-art results on FPV action recognition.

We introduce Ego4D, a massive-scale egocentric video dataset and benchmark suite. It offers 3,670 hours of dailylife activity video spanning hundreds of scenarios (household, outdoor, workplace, leisure, etc.) captured by 931 unique camera wearers from 74 worldwide locations and 9 different countries. The approach to collection is designed to uphold rigorous privacy and ethics standards, with consenting participants and robust de-identification procedures where relevant. Ego4D dramatically expands the volume of diverse egocentric video footage publicly available to the research community. Portions of the video are accompanied by audio, 3D meshes of the environment, eye gaze, stereo, and/or synchronized videos from multiple egocentric cameras at the same event. Furthermore, we present a host of new benchmark challenges centered around understanding the first-person visual experience in the past (querying an episodic memory), present (analyzing hand-object manipulation, audio-visual conversation, and social interactions), and future (forecasting activities). By publicly sharing this massive annotated dataset and benchmark suite, we aim to push the frontier of first-person perception. Project page: https://ego4d-data.org/

Based on deep learning technology, non-intrusive methods have received increasing attention for synthetic speech quality assessment since it does not need reference signals. Meanwhile, i-vector has been widely used in paralinguistic speech attribute recognition such as speaker and emotion recognition, but few studies have used it to estimate speech quality. In this paper, we propose a neural-network-based model that splices the deep features extracted by convolutional neural network (CNN) and i-vector on the time axis and uses Transformer encoder as time sequence model. To evaluate the proposed method, we improve the previous prediction models and conduct experiments on Voice Conversion Challenge (VCC) 2018 and 2016 dataset. Results show that i-vector contains information very related to the quality of synthetic speech and the proposed models that utilize i-vector and Transformer encoder highly increase the accuracy of MOSNet and MBNet on both utterance-level and system-level results.

We introduce a novel task of reconstructing a time series of second-person13D human body meshes from monocular egocentric videos. The unique viewpoint and rapid embodied camera motion of egocentric videos raise additional technical barriers for human body capture. To address those challenges, we propose a simple yet effective optimization-based approach that leverages 2D observations of the entire video sequence and human-scene interaction constraint to estimate second-person human poses, shapes, and global motion that are grounded on the 3D environment captured from the egocentric view. We conduct detailed ablation studies to validate our design choice. Moreover, we compare our method with the previous state-of-the-art method on human motion capture from monocular video, and show that our method estimates more accurate human-body poses and shapes under the challenging egocentric setting. In addition, we demonstrate that our approach produces more realistic human-scene interaction.