Paolo Rota

Guanglei Yang, Enrico Fini, Dan Xu et al.

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 (\method). In a nutshell, \method~is 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. The code is available at \url{https://github.com/ygjwd12345/UCD}.

Giacomo Zara, Subhankar Roy, Paolo Rota et al.

Open-set Unsupervised Video Domain Adaptation (OUVDA) 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 available.

Victor G. Turrisi da Costa, Giacomo Zara, Paolo Rota et al.

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$\leftrightarrow$UCF, as well as on Kinetics$\rightarrow$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.

Alessandro Conti, Enrico Fini, Massimiliano Mancini et al.

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.

Giacomo Zara, Alessandro Conti, Subhankar Roy et al.

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-V achieves significant improvement over state-of-the-art SFVUDA methods.

Hao Tang, Lei Ding, Songsong Wu et al.

Video processing and analysis have become an urgent task since a huge amount of videos (e.g., Youtube, Hulu) are uploaded online every day. The extraction of representative key frames from videos is very 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 Convolutional Neural Network (CNN) and Temporal Segment Density Peaks Clustering (TSDPC). The proposed TSDPC 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 (LSTM) is added on the top of the CNN to further elevate the performance of classification. Moreover, a weight fusion strategy of different input networks is presented to boost the 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.

Alessandro Conti, Paolo Rota, Yiming Wang et al.

Automatically understanding emotions from visual data is a fundamental task for human behaviour understanding. While models devised for Facial Expression Recognition (FER) have demonstrated excellent performances on many datasets, they often suffer from severe performance degradation when trained and tested on different datasets due to domain shift. In addition, as face images are considered highly sensitive data, the accessibility to large-scale datasets for model training is often denied. In this work, we tackle the above-mentioned problems by proposing the first Source-Free Unsupervised Domain Adaptation (SFUDA) method for FER. Our method exploits self-supervised pretraining to learn good feature representations from the target data and proposes a novel and robust cluster-level pseudo-labelling strategy that accounts for in-cluster statistics. We validate the effectiveness of our method in four adaptation setups, proving that it consistently outperforms existing SFUDA methods when applied to FER, and is on par with methods addressing FER in the UDA setting.

Giacomo Zara, Victor Guilherme Turrisi da Costa, Subhankar Roy et al.

In an effort to reduce annotation costs in action recognition, unsupervised video domain adaptation methods have been proposed that aim to adapt a predictive model from a labelled dataset (i.e., source domain) to an unlabelled dataset (i.e., target domain). In this work we address a more realistic scenario, called open-set video domain adaptation (OUVDA), where the target dataset contains "unknown" semantic categories that are not shared with the source. The challenge lies in aligning the shared classes of the two domains while separating the shared classes from the unknown ones. In this work we propose to address OUVDA with an unified contrastive learning framework that learns discriminative and well-clustered features. We also propose a video-oriented temporal contrastive loss that enables our method to better cluster the feature space by exploiting the freely available temporal information in video data. We show that discriminative feature space facilitates better separation of the unknown classes, and thereby allows us to use a simple similarity based score to identify them. We conduct thorough experimental evaluation on multiple OUVDA benchmarks and show the effectiveness of our proposed method against the prior art.

Tom Burgert, Leonard Hackel, Paolo Rota et al.

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.

Massimo Bosetti, Shibingfeng Zhang, Benedetta Liberatori et al.

Vision-language models (VLMs) have demonstrated remarkable performance across various visual tasks, leveraging joint learning of visual and textual representations. While these models excel in zero-shot image tasks, their application to zero-shot video action recognition (ZSVAR) remains challenging due to the dynamic and temporal nature of actions. Existing methods for ZS-VAR typically require extensive training on specific datasets, which can be resource-intensive and may introduce domain biases. In this work, we propose Text-Enhanced Action Recognition (TEAR), a simple approach to ZS-VAR that is training-free and does not require the availability of training data or extensive computational resources. Drawing inspiration from recent findings in vision and language literature, we utilize action descriptors for decomposition and contextual information to enhance zero-shot action recognition. Through experiments on UCF101, HMDB51, and Kinetics-600 datasets, we showcase the effectiveness and applicability of our proposed approach in addressing the challenges of ZS-VAR.

Benedetta Liberatori, Alessandro Conti, Lorenzo Vaquero et al.

Zero-shot temporal action localization (ZS-TAL) consists of classifying and localizing actions in untrimmed videos, where action classes are unseen at training time. Existing work uses Vision and Language Models (VLMs), taking advantage of their strong zero-shot transfer capabilities. Yet, these models face evident challenges with fine-grained action classification, making it difficult to directly use them to distinguish between the presence and absence of an action. Most current methods for ZS-TAL address these challenges by training models on large-scale video datasets, which require annotated data and often result in limited generalization performance. Recently, approaches discarding the use of labeled data have emerged as an alternative. Following this direction, we propose a novel approach, ``Textual Guidance for finer localization of actions in videos'' (TEGU), that compensates for the lack of supervision from training data by exploiting rich textual information derived from large language models and structured text extracted from captions. This additional linguistic context can improve fine-grained discrimination by providing richer cues about fine-grained action differences within videos. We validate the effectiveness of the proposed method by conducting experiments on the THUMOS14 and the ActivityNet-v1.3 datasets. Our results show that, by exploiting rich textual information for improved action localization, TEGU outperforms state-of-the-art ZS-TAL approaches that do not involve training

Yan Shu, Weichao Zeng, Fangmin Zhao et al.

Visual text is a crucial component in both document and scene images, conveying rich semantic information and attracting significant attention in the computer vision community. Beyond traditional tasks such as text detection and recognition, visual text processing has witnessed rapid advancements driven by the emergence of foundation models, including text image reconstruction and text image manipulation. Despite significant progress, challenges remain due to the unique properties that differentiate text from general objects. Effectively capturing and leveraging these distinct textual characteristics is essential for developing robust visual text processing models. In this survey, we present a comprehensive, multi-perspective analysis of recent advancements in visual text processing, focusing on two key questions: (1) What textual features are most suitable for different visual text processing tasks? (2) How can these distinctive text features be effectively incorporated into processing frameworks? Furthermore, we introduce VTPBench, a new benchmark that encompasses a broad range of visual text processing datasets. Leveraging the advanced visual quality assessment capabilities of multimodal large language models (MLLMs), we propose VTPScore, a novel evaluation metric designed to ensure fair and reliable evaluation. Our empirical study with more than 20 specific models reveals substantial room for improvement in the current techniques. Our aim is to establish this work as a fundamental resource that fosters future exploration and innovation in the dynamic field of visual text processing. The relevant repository is available at https://github.com/shuyansy/Visual-Text-Processing-survey.

Yan Shu, Bin Ren, Zhitong Xiong et al.

Vision-language models (VLMs) have shown promise in earth observation (EO), yet they struggle with tasks that require grounding complex spatial reasoning in precise pixel-level visual representations. To address this problem, we introduce TerraScope, a unified VLM that delivers pixel-grounded geospatial reasoning with two key capabilities: (1) modality-flexible reasoning: it handles single-modality inputs (optical or SAR) and adaptively fuses different modalities into the reasoning process when both are available; (2) multi-temporal reasoning: it integrates temporal sequences for change analysis across multiple time points. In addition, we curate Terra-CoT, a large-scale dataset containing 1 million samples with pixel-level masks embedded in reasoning chains across multiple sources. We also propose TerraScope-Bench, the first benchmark for pixel-grounded geospatial reasoning with six sub-tasks that evaluates both answer accuracy and mask quality to ensure authentic pixel-grounded reasoning. Experiments show that TerraScope significantly outperforms existing VLMs on pixel-grounded geospatial reasoning while providing interpretable visual evidence.

Shiyao Xu, Benedetta Liberatori, Gül Varol et al.

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.

Jiaqi Liu, Jichao Zhang, Paolo Rota et al.

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 DeepFashion dataset and enables flexible person image editing due to its generation consistency. The code is available at https://github.com/jqliu09/mcld.

Tom Burgert, Oliver Stoll, Paolo Rota et al.

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.

Guanglei Yang, Paolo Rota, Xavier Alameda-Pineda et al.

Convolutional neural networks have enabled major progresses in addressing pixel-level prediction tasks such as semantic segmentation, depth estimation, surface normal prediction and so on, benefiting from their powerful capabilities in visual representation learning. Typically, state of the art models integrate attention mechanisms for improved deep feature representations. Recently, some works have demonstrated the significance of learning and combining both spatial- and channelwise attentions for deep feature refinement. In this paper, weaim at effectively boosting previous approaches and propose a unified deep framework to jointly learn both spatial attention maps and channel attention vectors in a principled manner so as to structure the resulting attention tensors and model interactions between these two types of attentions. Specifically, we integrate the estimation and the interaction of the attentions within a probabilistic representation learning framework, leading to VarIational STructured Attention networks (VISTA-Net). We implement the inference rules within the neural network, thus allowing for end-to-end learning of the probabilistic and the CNN frontend parameters. As demonstrated by our extensive empirical evaluation on six large-scale datasets for dense visual prediction, VISTA-Net outperforms the state-of-the-art in multiple continuous and discrete prediction tasks, thus confirming the benefit of the proposed approach in joint structured spatial-channel attention estimation for deep representation learning. The code is available at https://github.com/ygjwd12345/VISTA-Net.

Benedetta Liberatori, Alessandro Conti, Paolo Rota et al.

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.

Alessandro Conti, Enrico Fini, Massimiliano Mancini et al.

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.

Xavier Alameda-Pineda, Angus Addlesee, Daniel Hernández García et al.

Despite the many recent achievements in developing and deploying social robotics, there are still many underexplored environments and applications for which systematic evaluation of such systems by end-users is necessary. While several robotic platforms have been used in gerontological healthcare, the question of whether or not a social interactive robot with multi-modal conversational capabilities will be useful and accepted in real-life facilities is yet to be answered. This paper is an attempt to partially answer this question, via two waves of experiments with patients and companions in a day-care gerontological facility in Paris with a full-sized humanoid robot endowed with social and conversational interaction capabilities. The software architecture, developed during the H2020 SPRING project, together with the experimental protocol, allowed us to evaluate the acceptability (AES) and usability (SUS) with more than 60 end-users. Overall, the users are receptive to this technology, especially when the robot perception and action skills are robust to environmental clutter and flexible to handle a plethora of different interactions.

Alessandro Conti, Massimiliano Mancini, Enrico Fini et al.

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.

Johann-Ludwig Herzog, Mathis Jürgen Adler, Leonard Hackel et al.

Vision-langugage models (VLMs) have shown strong performance in computer vision (CV), yet their performance on remote sensing (RS) data remains limited due to the lack of large-scale, multi-sensor RS image-text datasets with diverse textual annotations. Existing datasets predominantly include aerial Red-Green-Blue imagery, with short or weakly grounded captions, and provide limited diversity in annotation types. To address this limitation, we introduce BigEarthNet$.$txt, a large-scale, multi-sensor image-text dataset designed to advance instruction-driven image-text learning in Earth observation across multiple tasks. BigEarthNet$.$txt contains 464044 co-registered Sentinel-1 synthetic aperture radar and Sentinel-2 multispectral images with 9.6M text annotations, including: i) geographically anchored captions describing land-use/land-cover (LULC) classes, their spatial relations, and environmental context; ii) visual question answering pairs relevant for different tasks; and iii) referring expression detection instructions for bounding box prediction. Through a comparative statistical analysis, we demonstrate that BigEarthNet$.$txt surpasses existing RS image-text datasets in textual richness and annotation type variety. We further establish a manually-verified benchmark split to evaluate VLMs in RS and CV. The results show the limitations of these models on tasks that involve complex LULC classes, whereas fine-tuning using BigEarthNet$.$txt results in consistent performance gains across all considered tasks.

Benedetta Liberatori, Alessandro Conti, Lorenzo Vaquero et al.

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.

Yan Shu, Bin Ren, Zhitong Xiong et al.

Earth Observation (EO) data analysis is vital for monitoring environmental and human dynamics. Recent Multimodal Large Language Models (MLLMs) show potential in EO understanding but remain restricted to single-sensor inputs, overlooking the complementarity across heterogeneous modalities. We propose EarthMind, a unified vision-language framework that handles both single- and cross-sensor inputs via an innovative hierarchical cross-modal attention (ie, HCA) design. Specifically, HCA hierarchically captures visual relationships across sensors and aligns them with language queries, enabling adaptive fusion of optical and Synthetic Aperture Radar (SAR) features. To support cross-sensor learning, we curate FusionEO, a 30K-pair dataset with diverse annotations, and establish EarthMind-Bench, a 2,841-pair benchmark with expert annotations for perception and reasoning tasks. Extensive experiments show that EarthMind achieves state-of-the-art results on EarthMind-Bench and surpasses existing MLLMs on multiple EO benchmarks.

Alessandro Conti, Enrico Fini, Paolo Rota et al.

Assessing the capabilities of large multimodal models (LMMs) often requires the creation of ad-hoc evaluations. Currently, building new benchmarks requires tremendous amounts of manual work for each specific analysis. This makes the evaluation process tedious and costly. In this paper, we present APEx, Automatic Programming of Experiments, the first framework for automatic benchmarking of LMMs. Given a research question expressed in natural language, APEx leverages a large language model (LLM) and a library of pre-specified tools to generate a set of experiments for the model at hand, and progressively compile a scientific report. The report drives the testing procedure: based on the current status of the investigation, APEx chooses which experiments to perform and whether the results are sufficient to draw conclusions. Finally, the LLM refines the report, presenting the results to the user in natural language. Thanks to its modularity, our framework is flexible and extensible as new tools become available. Empirically, APEx reproduces the findings of existing studies while allowing for arbitrary analyses and hypothesis testing.

Gk Tejus, Giacomo Zara, Paolo Rota et al.

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.

Guanglei Yang, Enrico Fini, Dan Xu et al.

Over the past years, semantic segmentation, as many other tasks in computer vision, benefited from the progress in deep neural networks, resulting in significantly improved performance. However, deep architectures trained with gradient-based techniques suffer from catastrophic forgetting, which is the tendency to forget previously learned knowledge while learning new tasks. Aiming at devising strategies to counteract this effect, incremental learning approaches have gained popularity over the past years. However, the first incremental learning methods for semantic segmentation appeared only recently. While effective, these approaches do not account for a crucial aspect in pixel-level dense prediction problems, i.e. the role of attention mechanisms. To fill this gap, in this paper we introduce a novel attentive feature distillation approach to mitigate catastrophic forgetting while accounting for semantic spatial- and channel-level dependencies. Furthermore, we propose a {continual attentive fusion} structure, which takes advantage of the attention learned from the new and the old tasks while learning features for the new task. Finally, we also introduce a novel strategy to account for the background class in the distillation loss, thus preventing biased predictions. We demonstrate the effectiveness of our approach with an extensive evaluation on Pascal-VOC 2012 and ADE20K, setting a new state of the art.

Petru Soviany, Radu Tudor Ionescu, Paolo Rota et al.

Training machine learning models in a meaningful order, from the easy samples to the hard ones, using curriculum learning can provide performance improvements over the standard training approach based on random data shuffling, without any additional computational costs. Curriculum learning strategies have been successfully employed in all areas of machine learning, in a wide range of tasks. However, the necessity of finding a way to rank the samples from easy to hard, as well as the right pacing function for introducing more difficult data can limit the usage of the curriculum approaches. In this survey, we show how these limits have been tackled in the literature, and we present different curriculum learning instantiations for various tasks in machine learning. We construct a multi-perspective taxonomy of curriculum learning approaches by hand, considering various classification criteria. We further build a hierarchical tree of curriculum learning methods using an agglomerative clustering algorithm, linking the discovered clusters with our taxonomy. At the end, we provide some interesting directions for future work.

Jurandy Almeida, Cristiano Saltori, Paolo Rota et al.

Deep learning revolution happened thanks to the availability of a massive amount of labelled data which have contributed to the development of models with extraordinary inference capabilities. Despite the public availability of a large quantity of datasets, to address specific requirements it is often necessary to generate a new set of labelled data. Quite often, the production of labels is costly and sometimes it requires specific know-how to be fulfilled. In this work, we tackle a new problem named low-budget label query that consists in suggesting to the user a small (low budget) set of samples to be labelled, from a completely unlabelled dataset, with the final goal of maximizing the classification accuracy on that dataset. In this work we first improve an Unsupervised Domain Adaptation (UDA) method to better align source and target domains using consistency constraints, reaching the state of the art on a few UDA tasks. Finally, using the previously trained model as reference, we propose a simple yet effective selection method based on uniform sampling of the prediction consistency distribution, which is deterministic and steadily outperforms other baselines as well as competing models on a large variety of publicly available datasets.

Petru Soviany, Radu Tudor Ionescu, Paolo Rota et al.

Training (source) domain bias affects state-of-the-art object detectors, such as Faster R-CNN, when applied to new (target) domains. To alleviate this problem, researchers proposed various domain adaptation methods to improve object detection results in the cross-domain setting, e.g. by translating images with ground-truth labels from the source domain to the target domain using Cycle-GAN. On top of combining Cycle-GAN transformations and self-paced learning in a smart and efficient way, in this paper, we propose a novel self-paced algorithm that learns from easy to hard. Our method is simple and effective, without any overhead during inference. It uses only pseudo-labels for samples taken from the target domain, i.e. the domain adaptation is unsupervised. We conduct experiments on four cross-domain benchmarks, showing better results than the state of the art. We also perform an ablation study demonstrating the utility of each component in our framework. Additionally, we study the applicability of our framework to other object detectors. Furthermore, we compare our difficulty measure with other measures from the related literature, proving that it yields superior results and that it correlates well with the performance metric.

Marco Godi, Paolo Rota, Francesco Setti

Highlights in a sport video are usually referred as actions that stimulate excitement or attract attention of the audience. A big effort is spent in designing techniques which find automatically highlights, in order to automatize the otherwise manual editing process. Most of the state-of-the-art approaches try to solve the problem by training a classifier using the information extracted on the tv-like framing of players playing on the game pitch, learning to detect game actions which are labeled by human observers according to their perception of highlight. Obviously, this is a long and expensive work. In this paper, we reverse the paradigm: instead of looking at the gameplay, inferring what could be exciting for the audience, we directly analyze the audience behavior, which we assume is triggered by events happening during the game. We apply deep 3D Convolutional Neural Network (3D-CNN) to extract visual features from cropped video recordings of the supporters that are attending the event. Outputs of the crops belonging to the same frame are then accumulated to produce a value indicating the Highlight Likelihood (HL) which is then used to discriminate between positive (i.e. when a highlight occurs) and negative samples (i.e. standard play or time-outs). Experimental results on a public dataset of ice-hockey matches demonstrate the effectiveness of our method and promote further research in this new exciting direction.