Recent advances in 3D human motion and language integration have primarily focused on text-to-motion generation, leaving the task of motion understanding relatively unexplored. We introduce Dense Motion Captioning, a novel task that aims to temporally localize and caption actions within 3D human motion sequences. Current datasets fall short in providing detailed temporal annotations and predominantly consist of short sequences featuring few actions. To overcome these limitations, we present the Complex Motion Dataset (CompMo), the first large-scale dataset featuring richly annotated, complex motion sequences with precise temporal boundaries. Built through a carefully designed data generation pipeline, CompMo includes 60,000 motion sequences, each composed of multiple actions ranging from at least two to ten, accurately annotated with their temporal extents. We further present DEMO, a model that integrates a large language model with a simple motion adapter, trained to generate dense, temporally grounded captions. Our experiments show that DEMO substantially outperforms existing methods on CompMo as well as on adapted benchmarks, establishing a robust baseline for future research in 3D motion understanding and captioning.

Self-supervised learning (SSL) has become a powerful paradigm for learning from large, unlabeled datasets, particularly in computer vision (CV). However, applying SSL to multispectral remote sensing (RS) images presents unique challenges and opportunities due to the geographical and temporal variability of the data. In this paper, we introduce GeoRank, a novel regularization method for contrastive SSL that improves upon prior techniques by directly optimizing spherical distances to embed geographical relationships into the learned feature space. GeoRank outperforms or matches prior methods that integrate geographical metadata and consistently improves diverse contrastive SSL algorithms (e.g., BYOL, DINO). Beyond this, we present a systematic investigation of key adaptations of contrastive SSL for multispectral RS images, including the effectiveness of data augmentations, the impact of dataset cardinality and image size on performance, and the task dependency of temporal views. Code is available at https://github.com/tomburgert/georank.

Large vision-language models revolutionized image classification and semantic segmentation paradigms. However, they typically assume a pre-defined set of categories, or vocabulary, at test time for composing textual prompts. This assumption is impractical in scenarios with unknown or evolving semantic context. Here, we address this issue and introduce the Vocabulary-free Image Classification (VIC) task, which aims to assign a class from an unconstrained language-induced semantic space to an input image without needing a known vocabulary. VIC is challenging due to the vastness of the semantic space, which contains millions of concepts, including fine-grained categories. To address VIC, we propose Category Search from External Databases (CaSED), a training-free method that leverages a pre-trained vision-language model and an external database. CaSED first extracts the set of candidate categories from the most semantically similar captions in the database and then assigns the image to the best-matching candidate category according to the same vision-language model. Furthermore, we demonstrate that CaSED can be applied locally to generate a coarse segmentation mask that classifies image regions, introducing the task of Vocabulary-free Semantic Segmentation. CaSED and its variants outperform other more complex vision-language models, on classification and semantic segmentation benchmarks, while using much fewer parameters.

The Latent Diffusion Model (LDM) has demonstrated strong capabilities in high-resolution image generation and has been widely employed for Pose-Guided Person Image Synthesis (PGPIS), yielding promising results. However, the compression process of LDM often results in the deterioration of details, particularly in sensitive areas such as facial features and clothing textures. In this paper, we propose a Multi-focal Conditioned Latent Diffusion (MCLD) method to address these limitations by conditioning the model on disentangled, pose-invariant features from these sensitive regions. Our approach utilizes a multi-focal condition aggregation module, which effectively integrates facial identity and texture-specific information, enhancing the model’s ability to produce appearance realistic and identity-consistent images. Our method demonstrates consistent identity and appearance generation on the Deep-Fashion dataset and enables flexible person image editing due to its generation consistency. The code is available at https://github.com/jqliu09/mcld.

Traditional image classification requires a predefined list of semantic categories. In contrast, Large Multimodal Models (LMMs) can sidestep this requirement by classifying images directly using natural language (e.g., answering the prompt "What is the main object in the image?"). Despite this remarkable capability, most existing studies on LMM classification performance are surprisingly limited in scope, often assuming a closed-world setting with a predefined set of categories. In this work, we address this gap by thoroughly evaluating LMM classification performance in a truly open-world setting. We first formalize the task and introduce an evaluation protocol, defining various metrics to assess the alignment between predicted and ground truth classes. We then evaluate 13 models across 10 benchmarks, encompassing prototypical, non-prototypical, fine-grained, and very fine-grained classes, demonstrating the challenges LMMs face in this task. Further analyses based on the proposed metrics reveal the types of errors LMMs make, highlighting challenges related to granularity and fine-grained capabilities, showing how tailored prompting and reasoning can alleviate them.

The hypothesis that Convolutional Neural Networks (CNNs) are inherently texture-biased has shaped much of the discourse on feature use in deep learning. We revisit this hypothesis by examining limitations in the cue-conflict experiment by Geirhos et al. To address these limitations, we propose a domain-agnostic framework that quantifies feature reliance through systematic suppression of shape, texture, and color cues, avoiding the confounds of forced-choice conflicts. By evaluating humans and neural networks under controlled suppression conditions, we find that CNNs are not inherently texture-biased but predominantly rely on local shape features. Nonetheless, this reliance can be substantially mitigated through modern training strategies or architectures (ConvNeXt, ViTs). We further extend the analysis across computer vision, medical imaging, and remote sensing, revealing that reliance patterns differ systematically: computer vision models prioritize shape, medical imaging models emphasize color, and remote sensing models exhibit a stronger reliance on texture. Code is available at https://github.com/tomburgert/feature-reliance.

What does it mean for two videos to be similar? Videos may appear similar when judged by the actions they depict, yet entirely different if evaluated based on the locations where they were filmed. While humans naturally compare videos by taking different aspects into account, this ability has not been thoroughly studied and presents a challenge for models that often depend on broad global similarity scores. Large Multimodal Models (LMMs) with video understanding capabilities open new opportunities for leveraging natural language in comparative video tasks. We introduce Concept-based Video Similarity estimation (ConViS), a novel task that compares pairs of videos by computing interpretable similarity scores across a predefined set of key semantic concepts. ConViS allows for human-like reasoning about video similarity and enables new applications such as concept-conditioned video retrieval. To support this task, we also introduce ConViS-Bench, a new benchmark comprising carefully annotated video pairs spanning multiple domains. Each pair comes with concept-level similarity scores and textual descriptions of both differences and similarities. Additionally, we benchmark several state-of-the-art models on ConViS, providing insights into their alignment with human judgments. Our results reveal significant performance differences on ConViS, indicating that some concepts present greater challenges for estimating video similarity. We believe that ConViS-Bench will serve as a valuable resource for advancing research in language-driven video understanding.

Zero-Shot Temporal Action Localization (ZS-TAL) seeks to identify and locate actions in untrimmed videos unseen during training. Existing ZS-TAL methods involve fine-tuning a model on a large amount of annotated training data. While effective, training-based ZS-TAL approaches assume the availability of labeled data for supervised learning, which can be impractical in some applications. Furthermore, the training process naturally induces a domain bias into the learned model, which may adversely affect the model's generalization ability to arbitrary videos. These considerations prompt us to approach the ZS-TAL problem from a radically novel perspective, relaxing the requirement for training data. To this aim, we introduce a novel method that performs Test-Time adaptation for Temporal Action Localization (T3AL). In a nutshell, T3AL adapts a pre-trained Vision and Language Model (VLM). T3AL operates in three steps. First, a video-level pseudo-label of the action category is computed by aggregating information from the entire video. Then, action localization is performed adopting a novel procedure inspired by self-supervised learning. Finally, frame-level textual descriptions extracted with a state-of-the-art captioning model are employed for refining the action region proposals. We validate the effectiveness of T3AL by conducting experiments on the THUMOS14 and the ActivityNet-v1.3 datasets. Our results demonstrate that T3AL significantly outperforms zero-shot baselines based on state-of-the-art VLMs, confirming the benefit of a test-time adaptation approach.

Open-set Unsupervised Video Domain Adaptation (OU-VDA) deals with the task of adapting an action recognition model from a labelled source domain to an unlabelled target domain that contains “target-private” categories, which are present in the target but absent in the source. In this work we deviate from the prior work of training a specialized open-set classifier or weighted adversarial learning by proposing to use pre-trained Language and Vision Models (CLIP). The CLIP is well suited for OUVDA due to its rich representation and the zero-shot recognition capabilities. However, rejecting target-private instances with the CLIP's zero-shot protocol requires oracle knowledge about the target-private label names. To circumvent the impossibility of the knowledge of label names, we propose AutoLabel that automatically discovers and generates object-centric compositional candidate target-private class names. Despite its simplicity, we show that CLIP when equipped with AutoLabel can satisfactorily reject the target-private instances, thereby facilitating better alignment between the shared classes of the two domains. The code is available11https://github.com/gzaraunitn/autolabel.

Source-Free Video Unsupervised Domain Adaptation (SFVUDA) task consists in adapting an action recognition model, trained on a labelled source dataset, to an unlabelled target dataset, without accessing the actual source data. The previous approaches have attempted to address SFVUDA by leveraging self-supervision (e.g., enforcing temporal consistency) derived from the target data itself. In this work, we take an orthogonal approach by exploiting "web-supervision" from Large Language-Vision Models (LLVMs), driven by the rationale that LLVMs contain a rich world prior surprisingly robust to domain-shift. We showcase the unreasonable effectiveness of integrating LLVMs for SFVUDA by devising an intuitive and parameter-efficient method, which we name Domain Adaptation with Large Language-Vision models (DALL-V), that distills the world prior and complementary source model information into a student network tailored for the target. Despite the simplicity, DALL-V1achieves significant improvement over state-of-the-art SFVUDA methods.

In this paper we address the rotation synchronization problem, where the objective is to recover absolute rotations starting from pairwise ones, where the unknowns and the measures are represented as nodes and edges of a graph, respectively. This problem is an essential task for structure from motion and simultaneous localization and mapping. We focus on the formulation of synchronization via neural networks, which has only recently begun to be explored in the literature. Inspired by deep matrix completion, we express rotation synchronization in terms of matrix factorization with a deep neural network. Our formulation exhibits implicit regularization properties and, more importantly, is unsupervised, whereas previous deep approaches are supervised. Our experiments show that we achieve comparable accuracy to the closest competitors in most scenes, while working under weaker assumptions.

Recent advances in large vision-language models have revolutionized the image classification paradigm. Despite showing impressive zero-shot capabilities, a pre-defined set of categories, a.k.a. the vocabulary, is assumed at test time for composing the textual prompts. However, such assumption can be impractical when the semantic context is unknown and evolving. We thus formalize a novel task, termed as Vocabulary-free Image Classification (VIC), where we aim to assign to an input image a class that resides in an unconstrained language-induced semantic space, without the prerequisite of a known vocabulary. VIC is a challenging task as the semantic space is extremely large, containing millions of concepts, with hard-to-discriminate fine-grained categories. In this work, we first empirically verify that representing this semantic space by means of an external vision-language database is the most effective way to obtain semantically relevant content for classifying the image. We then propose Category Search from External Databases (CaSED), a method that exploits a pre-trained vision-language model and an external vision-language database to address VIC in a training-free manner. CaSED first extracts a set of candidate categories from captions retrieved from the database based on their semantic similarity to the image, and then assigns to the image the best matching candidate category according to the same vision-language model. Experiments on benchmark datasets validate that CaSED outperforms other complex vision-language frameworks, while being efficient with much fewer parameters, paving the way for future research in this direction.

A fundamental and challenging problem in deep learning is catastrophic forgetting, i.e., the tendency of neural networks to fail to preserve the knowledge acquired from old tasks when learning new tasks. This problem has been widely investigated in the research community and several Incremental Learning (IL) approaches have been proposed in the past years. While earlier works in computer vision have mostly focused on image classification and object detection, more recently some IL approaches for semantic segmentation have been introduced. These previous works showed that, despite its simplicity, knowledge distillation can be effectively employed to alleviate catastrophic forgetting. In this paper, we follow this research direction and, inspired by recent literature on contrastive learning, we propose a novel distillation framework, Uncertainty-aware Contrastive Distillation (UCD). In a nutshell, UCDis operated by introducing a novel distillation loss that takes into account all the images in a mini-batch, enforcing similarity between features associated to all the pixels from the same classes, and pulling apart those corresponding to pixels from different classes. In order to mitigate catastrophic forgetting, we contrast features of the new model with features extracted by a frozen model learned at the previous incremental step. Our experimental results demonstrate the advantage of the proposed distillation technique, which can be used in synergy with previous IL approaches, and leads to state-of-art performance on three commonly adopted benchmarks for incremental semantic segmentation.

International audience

Video processing and analysis have become an urgent task, as a huge amount of videos (e.g., YouTube, Hulu) are uploaded online every day. The extraction of representative key frames from videos is important in video processing and analysis since it greatly reduces computing resources and time. Although great progress has been made recently, large-scale video classification remains an open problem, as the existing methods have not well balanced the performance and efficiency simultaneously. To tackle this problem, this work presents an unsupervised method to retrieve the key frames, which combines the convolutional neural network and temporal segment density peaks clustering. The proposed temporal segment density peaks clustering is a generic and powerful framework, and it has two advantages compared with previous works. One is that it can calculate the number of key frames automatically. The other is that it can preserve the temporal information of the video. Thus, it improves the efficiency of video classification. Furthermore, a long short-term memory network is added on the top of the convolutional neural network to further elevate the performance of classification. Moreover, a weight fusion strategy of different input networks is presented to boost performance. By optimizing both video classification and key frame extraction simultaneously, we achieve better classification performance and higher efficiency. We evaluate our method on two popular datasets (i.e., HMDB51 and UCF101), and the experimental results consistently demonstrate that our strategy achieves competitive performance and efficiency compared with the state-of-the-art approaches.

Over the last few years, Unsupervised Domain Adaptation (UDA) techniques have acquired remarkable importance and popularity in computer vision. However, when compared to the extensive literature available for images, the field of videos is still relatively unexplored. On the other hand, the performance of a model in action recognition is heavily affected by domain shift. In this paper, we propose a simple and novel UDA approach for video action recognition. Our approach leverages recent advances on spatio-temporal transformers to build a robust source model that better generalises to the target domain. Furthermore, our architecture learns domain invariant features thanks to the introduction of a novel alignment loss term derived from the Information Bottleneck principle. We report results on two video action recognition benchmarks for UDA, showing state-of-the-art performance on HMDB ↔ UCF, as well as on Kinetics→NEC-Drone, which is more challenging. This demonstrates the effectiveness of our method in handling different levels of domain shift. The source code is available at https://github.com/vturrisi/UDAVT.

Unsupervised domain adaptation (UDA) methods have become very popular in computer vision. However, while several techniques have been proposed for images, much less attention has been devoted to videos. This paper introduces a novel UDA approach for action recognition from videos, inspired by recent literature on contrastive learning. In particular, we propose a novel two-headed deep architecture that simultaneously adopts cross-entropy and contrastive losses from different network branches to robustly learn a target classifier. Moreover, this work introduces a novel large-scale UDA dataset, Mixamo→Kinetics, which, to the best of our knowledge, is the first dataset that considers the domain shift arising when transferring knowledge from synthetic to real video sequences. Our extensive experimental evaluation conducted on three publicly available benchmarks and on our new Mixamo→Kinetics dataset demonstrate the effectiveness of our approach, which outperforms the current state-of-the-art methods. Code is available at https://github.com/vturrisi/CO2A.

Visible thermal person re-identification (VT-REID) is an important and challenging task in that 1) weak lighting environments are inevitably encountered in real-world settings and 2) the inter-modality discrepancy is serious. Most existing methods either aim at reducing the cross-modality gap in pixel- and feature-level or optimizing cross-modality network by metric learning techniques. However, few works have jointly considered these two aspects and studied their mutual benefits. In this paper, we design a novel framework to jointly bridge the modality gap in pixel- and feature-level without additional parameters, as well as reduce the inter- and intra-modalities variations by a center-guided metric learning constraint. Specifically, we introduce the Class-aware Modality Mix (CMM) to generate internal information of the two modalities for reducing the modality gap in pixel-level. In addition, we exploit the KL-divergence to further align modality distributions on feature-level. On the other hand, we propose an efficient Center-guided Metric Learning (CML) method for decreasing the discrepancy within the inter- and intra-modalities, by enforcing constraints on class centers and instances. Extensive experiments on two datasets show the mutual advantage of the proposed components and demonstrate the superiority of our method over the state of the art.

Deep learning (DL) has proved successful in medical imaging and, in the wake of the recent COVID-19 pandemic, some works have started to investigate DL-based solutions for the assisted diagnosis of lung diseases. While existing works focus on CT scans, this paper studies the application of DL techniques for the analysis of lung ultrasonography (LUS) images. Specifically, we present a novel fully-annotated dataset of LUS images collected from several Italian hospitals, with labels indicating the degree of disease severity at a frame-level, video-level, and pixel-level (segmentation masks). Leveraging these data, we introduce several deep models that address relevant tasks for the automatic analysis of LUS images. In particular, we present a novel deep network, derived from Spatial Transformer Networks, which simultaneously predicts the disease severity score associated to a input frame and provides localization of pathological artefacts in a weakly-supervised way. Furthermore, we introduce a new method based on uninorms for effective frame score aggregation at a video-level. Finally, we benchmark state of the art deep models for estimating pixel-level segmentations of COVID-19 imaging biomarkers. Experiments on the proposed dataset demonstrate satisfactory results on all the considered tasks, paving the way to future research on DL for the assisted diagnosis of COVID-19 from LUS data.

The pervasive availability of the Internet, coupled with the development of increasingly powerful technologies, has led digital images to be the primary source of visual information in nowadays society. However, their reliability as a true representation of reality cannot be taken for granted, due to the affordable powerful graphics editing softwares that can easily alter the original content, leaving no visual trace of any modification on the image making them potentially dangerous. This motivates developing technological solutions able to detect media manipulations without a prior knowledge or extra information regarding the given image. At the same time, the huge amount of available data has also led to tremendous advances of data-hungry learning models, which have already demonstrated in last few years to be successful in image classification. In this work we propose a deep learning approach for tampered image classification. To our best knowledge, this the first attempt to use the deep learning paradigm in an image forensic scenario. In particular, we propose a new blind deep learning approach based on Convolutional Neural Networks (CNN) able to learn invisible discriminative artifacts from manipulated images that can be exploited to automatically discriminate between forged and authentic images. The proposed approach not only detects forged images but it can be extended to localize the tampered regions within the image. This method outperforms the state-of-the-art in terms of accuracy on CASIA TIDE v2.0 dataset. The capability of automatically crafting discriminant features can lead to surprising results. For instance, detecting image compression filters used to create the dataset. This argument is also discussed within this paper.

Minimal Residual Disease (MRD) is a powerful risk-stratification marker in the treatment of primary and relapsed childhood Acute Lymphoblastic Leukemia. Flow Cytometry is a fast and sensitive method to detect MRD, however, the interpretation of these multi-parametric data demands intensive operator training and experience.

The topic of crowd modeling in computer vision usually assumes a single generic typology of crowd, which is very simplistic. In this paper we adopt a taxonomy that is widely accepted in sociology, focusing on a particular category, the spectator crowd, which is formed by people “interested in watching something specific that they came to see” [6]. This can be found at the stadiums, amphitheaters, cinema, etc. In particular, we propose a novel dataset, the Spectators Hockey (S-HOCK), which deals with 4 hockey matches during an international tournament. In the dataset, a massive annotation has been carried out, focusing on the spectators at different levels of details: at a higher level, people have been labeled depending on the team they are supporting and the fact that they know the people close to them; going to the lower levels, standard pose information has been considered (regarding the head, the body) but also fine grained actions such as hands on hips, clapping hands etc. The labeling focused on the game field also, permitting to relate what is going on in the match with the crowd behavior. This brought to more than 100 millions of annotations, useful for standard applications as people counting and head pose estimation but also for novel tasks as spectator categorization. For all of these we provide protocols and baseline results, encouraging further research.

This paper proposes a method to detect and localize dyadic human interactions in real videos. The idea stems from the significant difference between an action performed by a single subject and an interaction between two persons. In the first case all the visual information is concentrated on the subject, while in the latter case the action of a person is related to the interacting person's attitude, following an action/reaction principle. This kind of behavior is significant especially in natural and real scenarios, in which people are moving freely without the awareness of being recorded. To highlight these features and provide researchers with a common ground for comparisons, we have collected and annotated a new dataset, retrieving from YouTube 30 different videos of a specific type of interaction, namely urban fight situations. The proposed dataset is one of the most challenging annotated video collection concerning dyadic interactions, due to the intrinsic intra-class variability characterizing real fights. In addition, we provide an extensive experimental analysis on this dataset and we demonstrate that the visual information extracted in the area associated to the interpersonal space plays a fundamental role in detecting fights.

In this paper, we propose a framework to recognize complex human interactions. First, we adopt trajectories to represent human motion in a video. Then, the extracted trajectories are clustered into different groups (named as local motion patterns) using the coherent filtering algorithm. As trajectories within the same group exhibit similar motion properties (i.e., velocity, direction), we adopt the histogram of large-displacement optical flow (denoted as HO-LDOF) as the group motion feature vector. Thus, each video can be briefly represented by a collection of local motion patterns that are described by the HO-LDOF. Finally, classification is achieved using the citation-KNN, which is a typical multiple-instance-learning algorithm. Experimental results on the TV human interaction dataset and the UT human interaction dataset demonstrate the applicability of our method.