Fu-Jen Chu

Kristen Grauman, Andrew Westbury, Lorenzo Torresani et al. · cmu, gatech

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. Project page: http://ego-exo4d-data.org/

Md Mohaiminul Islam, Tushar Nagarajan, Huiyu Wang et al.

Goal-oriented planning, or anticipating a series of actions that transition an agent from its current state to a predefined objective, is crucial for developing intelligent assistants aiding users in daily procedural tasks. The problem presents significant challenges due to the need for comprehensive knowledge of temporal and hierarchical task structures, as well as strong capabilities in reasoning and planning. To achieve this, prior work typically relies on extensive training on the target dataset, which often results in significant dataset bias and a lack of generalization to unseen tasks. In this work, we introduce VidAssist, an integrated framework designed for zero/few-shot goal-oriented planning in instructional videos. VidAssist leverages large language models (LLMs) as both the knowledge base and the assessment tool for generating and evaluating action plans, thus overcoming the challenges of acquiring procedural knowledge from small-scale, low-diversity datasets. Moreover, VidAssist employs a breadth-first search algorithm for optimal plan generation, in which a composite of value functions designed for goal-oriented planning is utilized to assess the predicted actions at each step. Extensive experiments demonstrate that VidAssist offers a unified framework for different goal-oriented planning setups, e.g., visual planning for assistance (VPA) and procedural planning (PP), and achieves remarkable performance in zero-shot and few-shot setups. Specifically, our few-shot model outperforms the prior fully supervised state-of-the-art method by +7.7% in VPA and +4.81% PP task on the COIN dataset while predicting 4 future actions. Code, and models are publicly available at https://sites.google.com/view/vidassist.

Zi-Yi Dou, Xitong Yang, Tushar Nagarajan et al.

We present EMBED (Egocentric Models Built with Exocentric Data), a method designed to transform exocentric video-language data for egocentric video representation learning. Large-scale exocentric data covers diverse activities with significant potential for egocentric learning, but inherent disparities between egocentric and exocentric data pose challenges in utilizing one view for the other seamlessly. Egocentric videos predominantly feature close-up hand-object interactions, whereas exocentric videos offer a broader perspective on human activities. Additionally, narratives in egocentric datasets are typically more action-centric and closely linked with the visual content, in contrast to the narrative styles found in exocentric datasets. To address these challenges, we employ a data transformation framework to adapt exocentric data for egocentric training, focusing on identifying specific video clips that emphasize hand-object interactions and transforming narration styles to align with egocentric perspectives. By applying both vision and language style transfer, our framework creates a new egocentric dataset derived from exocentric video-language data. Through extensive evaluations, we demonstrate the effectiveness of EMBED, achieving state-of-the-art results across various egocentric downstream tasks, including an absolute improvement of 4.7% on the Epic-Kitchens-100 multi-instance retrieval and 6.2% on the EGTEA classification benchmarks in zero-shot settings. Furthermore, EMBED enables egocentric video-language models to perform competitively in exocentric tasks. Finally, we showcase EMBED's application across various exocentric datasets, exhibiting strong generalization capabilities when applied to different exocentric datasets.

SAM 3D Team, Xingyu Chen, Fu-Jen Chu et al.

We present SAM 3D, a generative model for visually grounded 3D object reconstruction, predicting geometry, texture, and layout from a single image. SAM 3D excels in natural images, where occlusion and scene clutter are common and visual recognition cues from context play a larger role. We achieve this with a human- and model-in-the-loop pipeline for annotating object shape, texture, and pose, providing visually grounded 3D reconstruction data at unprecedented scale. We learn from this data in a modern, multi-stage training framework that combines synthetic pretraining with real-world alignment, breaking the 3D "data barrier". We obtain significant gains over recent work, with at least a 5:1 win rate in human preference tests on real-world objects and scenes. We will release our code and model weights, an online demo, and a new challenging benchmark for in-the-wild 3D object reconstruction.

Runsen Xu, Weiyao Wang, Hao Tang et al.

Multi-modal large language models (MLLMs) have rapidly advanced in visual tasks, yet their spatial understanding remains limited to single images, leaving them ill-suited for robotics and other real-world applications that require multi-frame reasoning. In this paper, we propose a framework to equip MLLMs with robust multi-frame spatial understanding by integrating depth perception, visual correspondence, and dynamic perception. Central to our approach is the MultiSPA dataset, a novel, large-scale collection of more than 27 million samples spanning diverse 3D and 4D scenes. Alongside MultiSPA, we introduce a comprehensive benchmark that tests a wide spectrum of spatial tasks under uniform metrics. Our resulting model, Multi-SpatialMLLM, achieves significant gains over baselines and proprietary systems, demonstrating scalable, generalizable multi-frame reasoning. We further observe multi-task benefits and early indications of emergent capabilities in challenging scenarios, and showcase how our model can serve as a multi-frame reward annotator for robotics.

Mingzhen Huang, Fu-Jen Chu, Bugra Tekin et al.

We introduce HOIGPT, a token-based generative method that unifies 3D hand-object interactions (HOI) perception and generation, offering the first comprehensive solution for captioning and generating high-quality 3D HOI sequences from a diverse range of conditional signals (\eg text, objects, partial sequences). At its core, HOIGPT utilizes a large language model to predict the bidrectional transformation between HOI sequences and natural language descriptions. Given text inputs, HOIGPT generates a sequence of hand and object meshes; given (partial) HOI sequences, HOIGPT generates text descriptions and completes the sequences. To facilitate HOI understanding with a large language model, this paper introduces two key innovations: (1) a novel physically grounded HOI tokenizer, the hand-object decomposed VQ-VAE, for discretizing HOI sequences, and (2) a motion-aware language model trained to process and generate both text and HOI tokens. Extensive experiments demonstrate that HOIGPT sets new state-of-the-art performance on both text generation (+2.01% R Precision) and HOI generation (-2.56 FID) across multiple tasks and benchmarks.

Nikhil Mehta, Kevin J Liang, Jing Huang et al.

Out-of-distribution (OOD) detection is an important topic for real-world machine learning systems, but settings with limited in-distribution samples have been underexplored. Such few-shot OOD settings are challenging, as models have scarce opportunities to learn the data distribution before being tasked with identifying OOD samples. Indeed, we demonstrate that recent state-of-the-art OOD methods fail to outperform simple baselines in the few-shot setting. We thus propose a hypernetwork framework called HyperMix, using Mixup on the generated classifier parameters, as well as a natural out-of-episode outlier exposure technique that does not require an additional outlier dataset. We conduct experiments on CIFAR-FS and MiniImageNet, significantly outperforming other OOD methods in the few-shot regime.

Yunzhi Lin, Chao Tang, Fu-Jen Chu et al.

Shape informs how an object should be grasped, both in terms of where and how. As such, this paper describes a segmentation-based architecture for decomposing objects sensed with a depth camera into multiple primitive shapes, along with a post-processing pipeline for robotic grasping. Segmentation employs a deep network, called PS-CNN, trained on synthetic data with 6 classes of primitive shapes and generated using a simulation engine. Each primitive shape is designed with parametrized grasp families, permitting the pipeline to identify multiple grasp candidates per shape region. The grasps are rank ordered, with the first feasible one chosen for execution. For task-free grasping of individual objects, the method achieves a 94.2% success rate placing it amongst the top performing grasp methods when compared to top-down and SE(3)-based approaches. Additional tests involving variable viewpoints and clutter demonstrate robustness to setup. For task-oriented grasping, PS-CNN achieves a 93.0% success rate. Overall, the outcomes support the hypothesis that explicitly encoding shape primitives within a grasping pipeline should boost grasping performance, including task-free and task-relevant grasp prediction.

Ruinian Xu, Fu-Jen Chu, Patricio A. Vela

Contemporary grasp detection approaches employ deep learning to achieve robustness to sensor and object model uncertainty. The two dominant approaches design either grasp-quality scoring or anchor-based grasp recognition networks. This paper presents a different approach to grasp detection by treating it as keypoint detection in image-space. The deep network detects each grasp candidate as a pair of keypoints, convertible to the grasp representationg = {x, y, w, θ} T , rather than a triplet or quartet of corner points. Decreasing the detection difficulty by grouping keypoints into pairs boosts performance. To promote capturing dependencies between keypoints, a non-local module is incorporated into the network design. A final filtering strategy based on discrete and continuous orientation prediction removes false correspondences and further improves grasp detection performance. GKNet, the approach presented here, achieves a good balance between accuracy and speed on the Cornell and the abridged Jacquard datasets (96.9% and 98.39% at 41.67 and 23.26 fps). Follow-up experiments on a manipulator evaluate GKNet using 4 types of grasping experiments reflecting different nuisance sources: static grasping, dynamic grasping, grasping at varied camera angles, and bin picking. GKNet outperforms reference baselines in static and dynamic grasping experiments while showing robustness to varied camera viewpoints and moderate clutter. The results confirm the hypothesis that grasp keypoints are an effective output representation for deep grasp networks that provide robustness to expected nuisance factors.

Yunzhi Lin, Chao Tang, Fu-Jen Chu et al.

A segmentation-based architecture is proposed to decompose objects into multiple primitive shapes from monocular depth input for robotic manipulation. The backbone deep network is trained on synthetic data with 6 classes of primitive shapes generated by a simulation engine. Each primitive shape is designed with parametrized grasp families, permitting the pipeline to identify multiple grasp candidates per shape primitive region. The grasps are priority ordered via proposed ranking algorithm, with the first feasible one chosen for execution. On task-free grasping of individual objects, the method achieves a 94% success rate. On task-oriented grasping, it achieves a 76% success rate. Overall, the method supports the hypothesis that shape primitives can support task-free and task-relevant grasp prediction.

Fu-Jen Chu, Ruinian Xu, Chao Tang et al.

Affordance information about a scene provides important clues as to what actions may be executed in pursuit of meeting a specified goal state. Thus, integrating affordance-based reasoning into symbolic action plannning pipelines would enhance the flexibility of robot manipulation. Unfortunately, the top performing affordance recognition methods use object category priors to boost the accuracy of affordance detection and segmentation. Object priors limit generalization to unknown object categories. This paper describes an affordance recognition pipeline based on a category-agnostic region proposal network for proposing instance regions of an image across categories. To guide affordance learning in the absence of category priors, the training process includes the auxiliary task of explicitly inferencing existing affordances within a proposal. Secondly, a self-attention mechanism trained to interpret each proposal learns to capture rich contextual dependencies through the region. Visual benchmarking shows that the trained network, called AffContext, reduces the performance gap between object-agnostic and object-informed affordance recognition. AffContext is linked to the Planning Domain Definition Language (PDDL) with an augmented state keeper for action planning across temporally spaced goal-oriented tasks. Manipulation experiments show that AffContext can successfully parse scene content to seed a symbolic planner problem specification, whose execution completes the target task. Additionally, task-oriented grasping for cutting and pounding actions demonstrate the exploitation of multiple affordances for a given object to complete specified tasks.

Fu-Jen Chu, Ruinian Xu, Patricio A. Vela

A deep learning architecture is proposed to predict graspable locations for robotic manipulation. It considers situations where no, one, or multiple object(s) are seen. By defining the learning problem to be classification with null hypothesis competition instead of regression, the deep neural network with RGB-D image input predicts multiple grasp candidates for a single object or multiple objects, in a single shot. The method outperforms state-of-the-art approaches on the Cornell dataset with 96.0% and 96.1% accuracy on image-wise and object- wise splits, respectively. Evaluation on a multi-object dataset illustrates the generalization capability of the architecture. Grasping experiments achieve 96.0% grasp localization and 88.0% grasping success rates on a test set of household objects. The real-time process takes less than .25 s from image to plan.

Fu-Jen Chu, Ruinian Xu, Zhenxuan Zhang et al.

A human-in-the-loop system is proposed to enable collaborative manipulation tasks for person with physical disabilities. Studies show that the cognitive burden of subject reduces with increased autonomy of assistive system. Our framework obtains high-level intent from the user to specify manipulation tasks. The system processes sensor input to interpret the user's environment. Augmented reality glasses provide ego-centric visual feedback of the interpretation and summarize robot affordances on a menu. A tongue drive system serves as the input modality for triggering a robotic arm to execute the tasks. Assistance experiments compare the system to Cartesian control and to state-of-the-art approaches. Our system achieves competitive results with faster completion time by simplifying manipulation tasks.

Justin S. Smith, Jin-Ha Hwang, Fu-Jen Chu et al.

Recent applications of deep learning to navigation have generated end-to-end navigation solutions whereby visual sensor input is mapped to control signals or to motion primitives. The resulting visual navigation strategies work very well at collision avoidance and have performance that matches traditional reactive navigation algorithms while operating in real-time. It is accepted that these solutions cannot provide the same level of performance as a global planner. However, it is less clear how such end-to-end systems should be integrated into a full navigation pipeline. We evaluate the typical end-to-end solution within a full navigation pipeline in order to expose its weaknesses. Doing so illuminates how to better integrate deep learning methods into the navigation pipeline. In particular, we show that they are an efficient means to provide informed samples for sample-based planners. Controlled simulations with comparison against traditional planners show that the number of samples can be reduced by an order of magnitude while preserving navigation performance. Implementation on a mobile robot matches the simulated performance outcomes.

Pin-Yu Chen, Shin-Ming Cheng, Pai-Shun Ting et al.

Mobile sensing is an emerging technology that utilizes agent-participatory data for decision making or state estimation, including multimedia applications. This article investigates the structure of mobile sensing schemes and introduces crowdsourcing methods for mobile sensing. Inspired by social network, one can establish trust among participatory agents to leverage the wisdom of crowds for mobile sensing. A prototype of social network inspired mobile multimedia and sensing application is presented for illustrative purpose. Numerical experiments on real-world datasets show improved performance of mobile sensing via crowdsourcing. Challenges for mobile sensing with respect to Internet layers are discussed.

Pin-Yu Chen, Chia-Wei Lien, Fu-Jen Chu et al.

Crowdsourcing utilizes the wisdom of crowds for collective classification via information (e.g., labels of an item) provided by labelers. Current crowdsourcing algorithms are mainly unsupervised methods that are unaware of the quality of crowdsourced data. In this paper, we propose a supervised collective classification algorithm that aims to identify reliable labelers from the training data (e.g., items with known labels). The reliability (i.e., weighting factor) of each labeler is determined via a saddle point algorithm. The results on several crowdsourced data show that supervised methods can achieve better classification accuracy than unsupervised methods, and our proposed method outperforms other algorithms.