We introduce ED-VTG, a method for fine-grained video temporal grounding utilizing multi-modal large language models. Our approach harnesses the capabilities of multimodal LLMs to jointly process text and video, in order to effectively localize natural language queries in videos through a two-stage process. Rather than being directly grounded, language queries are initially transformed into enriched sentences that incorporate missing details and cues to aid in grounding. In the second stage, these enriched queries are grounded, using a lightweight decoder, which specializes at predicting accurate boundaries conditioned on contextualized representations of the enriched queries. To mitigate noise and reduce the impact of hallucinations, our model is trained with a multiple-instance-learning objective that dynamically selects the optimal version of the query for each training sample. We demonstrate state-of-the-art results across various benchmarks in temporal video grounding and paragraph grounding settings. Experiments reveal that our method significantly outperforms all previously proposed LLM-based temporal grounding approaches and is either superior or comparable to specialized models, while maintaining a clear advantage against them in zero-shot evaluation scenarios.

Vision-language models are integral to computer vision research, yet many high-performing models remain closed-source, obscuring their data, design and training recipe. The research community has responded by using distillation from black-box models to label training data, achieving strong benchmark results, at the cost of measurable scientific progress. However, without knowing the details of the teacher model and its data sources, scientific progress remains difficult to measure. In this paper, we study building a Perception Language Model (PLM) in a fully open and reproducible framework for transparent research in image and video understanding. We analyze standard training pipelines without distillation from proprietary models and explore large-scale synthetic data to identify critical data gaps, particularly in detailed video understanding. To bridge these gaps, we release 2.8M human-labeled instances of fine-grained video question-answer pairs and spatio-temporally grounded video captions. Additionally, we introduce PLM–VideoBench, a suite for evaluating challenging video understanding tasks focusing on the ability to reason about ''what'', ''where'', ''when'', and ''how'' of a video. We make our work fully reproducible by providing data, training recipes, code & models.

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 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.

Egocentric videos provide a first-person perspective of the wearer's activities, involving simultaneous interactions with multiple objects. In this work, we propose the task of weakly-supervised Narration-based Video Object Segmentation (NVOS). Given an egocentric video clip and a narration of the wearer's activities, our aim is to segment object instances mentioned in the narration, with-out using any spatial annotations during training. Existing weakly-supervised video object grounding methods typ-ically yield bounding boxes for referred objects. In contrast, we propose ROSA, a weakly-supervised pixel-level grounding framework learning alignments between referred objects and segmentation mask proposals. Our model harnesses vision-language models pre-trained on image-text pairs to embed region masks and object phrases. During training, we combine (a) a video-narration contrastive loss that implicitly supervises the alignment between regions and phrases, and (b) a region-phrase contrastive loss based on inferred latent alignments. To address the lack of annotated NVOS datasets in egocentric videos, we create a new evaluation benchmark, VISOR-NVOS, leveraging existing annotations of segmentation masks from VISOR alongside 14.6k newly-collected, object-based video clip narrations. Our approach achieves state-of-the-art zero-shot pixel-level grounding performance compared to strong baselines under similar supervision. Additionally, we demonstrate generalization capabilities for zero-shot video object grounding on YouCook2, a third-person instructional video dataset.

In this paper we present an approach for localizing steps of procedural activities in narrated how-to videos. To deal with the scarcity of labeled data at scale, we source the step descriptions from a language knowledge base (wikiHow) containing instructional articles for a large variety of procedural tasks. Without any form of manual supervision, our model learns to temporally ground the steps of procedural articles in how-to videos by matching three modalities: frames, narrations, and step descriptions. Specifically, our method aligns steps to video by fusing information from two distinct pathways: i) direct alignment of step descriptions to frames, ii) indirect alignment obtained by composing steps-to-narrations with narrations-to-video correspondences. Notably, our approach performs global temporal grounding of all steps in an article at once by exploiting order information, and is trained with step pseudo-labels which are iteratively refined and aggressively filtered. In order to validate our model we introduce a new benchmark – HT-Step – obtained by manually annotating a 124-hour subset of HowTo100M1with steps sourced from wikiHow articles. Experiments on this benchmark as well as zero-shot evaluations on CrossTask demonstrate that our multi-modality alignment yields dramatic gains over several baselines and prior works. Finally, we show that our inner module for matching narration-to-video outperforms by a large margin the state of the art on the HTM-Align narration-video alignment benchmark.

We introduce HT-Step, a large-scale dataset containing temporal annotations of instructional article steps in cooking videos. It includes 122k segment-level annotations over 20k narrated videos (approximately 2.3k hours) of the HowTo100M dataset.Each annotation provides a temporal interval, and a categorical step label from a taxonomy of 4,958 unique steps automatically mined from wikiHow articles which include rich descriptions of each step.Our dataset significantly surpasses existing labeled step datasets in terms of scale, number of tasks, and richness of natural language step descriptions. Based on these annotations, we introduce a strongly supervised benchmark for aligning instructional articles with how-to videos and present a comprehensive evaluation of baseline methods for this task.By publicly releasing these annotations and defining rigorous evaluation protocols and metrics,we hope to significantly accelerate research in the field of procedural activity understanding.

Given weak supervision from image- or video-caption pairs, we address the problem of grounding (localizing) each object word of a ground-truth or generated sentence describing a visual input. Recent weakly-supervised approaches leverage region proposals and ground words based on the region attention coefficients of captioning models. To predict each next word in the sentence they attend over regions using a summary of the previous words as a query, and then ground the word by selecting the most attended regions. However, this leads to sub-optimal grounding, since attention coefficients are computed without taking into account the word that needs to be localized. To address this shortcoming, we propose a novel Grounded Visual Description Conditional Variational Autoencoder (GVD-CVAE) and leverage its latent variables for grounding. In particular, we introduce a discrete random variable that models each word-to-region alignment, and learn its approximate posterior distribution given the full sentence. Experiments on challenging image and video datasets (Flickr30k Entities, YouCook2, ActivityNet Entities) validate the effectiveness of our conditional generative model, showing that it can substantially outperform soft-attention-based baselines in grounding.

Activity recognition in children and infants is important in applications such as safety monitoring, behavior assessment, and child-robot interaction, among others. However, it differs from activity recognition in adults not only because body poses and proportions are different, but also because of the way in which actions are performed. This paper addresses the problem of infant action classification in challenging conditions. The actions are performed in a pediatric rehabilitation environment in which not only infants but also robots and adults are present, with the infant being one of the smallest actors in the scene. We propose a multiview action classification system based on Faster R-CNN and LSTM networks, which fuses information from different views by using learnable fusion coefficients derived from detection confidence scores. The proposed system is view-independent, learns features that are close to view-invariant, and can handle new or missing views at test time. Our approach outperforms the state-of-the-art baseline model for a small dataset (2 subjects, 10-24 months old) by 11.4% in terms of average classification accuracy in four classes (crawl, sit, stand and walk). Moreover, experiments in an extended dataset (6 subjects, 8-24 months old) show that the proposed fusion strategy outperforms all the alternative fusion methods studied.

Fine-grained action segmentation and recognition is an important yet challenging task. Given a long, untrimmed sequence of kinematic data, the task is to classify the action at each time frame and segment the time series into the correct sequence of actions. In this paper, we propose a novel framework that combines a temporal Conditional Random Field (CRF) model with a powerful frame-level representation based on discriminative sparse coding. We introduce an end-to-end algorithm for jointly learning the weights of the CRF model, which include action classification and action transition costs, as well as an overcomplete dictionary of mid-level action primitives. This results in a CRF model that is driven by sparse coding features obtained using a discriminative dictionary that is shared among different actions and adapted to the task of structured output learning. We evaluate our method on three surgical tasks using kinematic data from the JIGSAWS dataset, as well as on a food preparation task using accelerometer data from the 50 Salads dataset. Our results show that the proposed method performs on par or better than state-of-the-art methods.

We propose a new approach to action classification in video, which uses deep appearance and motion features extracted from spatio-temporal volumes defined along body part trajectories to learn mid-level classifiers called deep moving poselets. A deep moving poselet is a classifier that captures a characteristic body part configuration, with a specific appearance and undergoing a specific movement. By having this mid-level representation of a body part be shared across action classes and by learning it jointly with action classifiers, we obtain a representation that is interpretable, shared and discriminative. In addition, by using sparsity-inducing norms to regularize action classifiers, we can reduce the number of deep moving poselets used by each class without hurting performance. Experiments show that the proposed method achieves state-of-the-art performance on the popular and challenging sub-JHMDB and MSR Daily Activity datasets.