Giuseppe Averta

Andrea Zenotto, Simone Alberto Peirone, Francesca Pistilli et al.

Procedural activities follow well-defined structures: whether we consider a cooking recipe or a mechanic repairing a car, these activities naturally decompose in a hierarchy of steps and sub-steps. Traditional approaches for step grounding require extensive annotations and scale poorly. Instead, we argue that such hierarchical structure can emerge naturally from uncurated videos of human activities through recurring patterns of co-occurring actions and activities. Our approach builds on HiERO, a weakly-supervised representation learning approach that maps close in the feature space actions that are functionally related to each other, leveraging only fine-grained action-level narrations. In this feature space, procedure steps can be detected by a simple clustering, with no additional task-specific fine-tuning. For the Ego4D Step Grounding challenge, we augment this approach by ensuring fine and coarse level agreement in step assignments, enforcing strict temporal monotonicity of the grounded steps and post-processing the detected steps to reduce the impact of noisy predictions. We call this approach HiERO-StepG and it achieves 56.27 % on the R@1 (IoU = 0.3) metric on the global leaderboard at submission time, ranking second while being completely zero-shot and not requiring procedure-specific annotations. Project page: https://github.com/andreazenotto/HiERO-StepG.

Niccolò Cavagnero, Luca Robbiano, Barbara Caputo et al.

In the last decade, most research in Machine Learning contributed to the improvement of existing models, with the aim of increasing the performance of neural networks for the solution of a variety of different tasks. However, such advancements often come at the cost of an increase of model memory and computational requirements. This represents a significant limitation for the deployability of research output in realistic settings, where the cost, the energy consumption, and the complexity of the framework play a crucial role. To solve this issue, the designer should search for models that maximise the performance while limiting its footprint. Typical approaches to reach this goal rely either on manual procedures, which cannot guarantee the optimality of the final design, or upon Neural Architecture Search algorithms to automatise the process, at the expenses of extremely high computational time. This paper provides a solution for the fast identification of a neural network that maximises the model accuracy while preserving size and computational constraints typical of tiny devices. Our approach, named FreeREA, is a custom cell-based evolution NAS algorithm that exploits an optimised combination of training-free metrics to rank architectures during the search, thus without need of model training. Our experiments, carried out on the common benchmarks NAS-Bench-101 and NATS-Bench, demonstrate that i) FreeREA is a fast, efficient, and effective search method for models automatic design; ii) it outperforms State of the Art training-based and training-free techniques in all the datasets and benchmarks considered, and iii) it can easily generalise to constrained scenarios, representing a competitive solution for fast Neural Architecture Search in generic constrained applications. The code is available at \url{https://github.com/NiccoloCavagnero/FreeREA}.

Gabriele Tiboni, Karol Arndt, Giuseppe Averta et al.

Physics simulators have shown great promise for conveniently learning reinforcement learning policies in safe, unconstrained environments. However, transferring the acquired knowledge to the real world can be challenging due to the reality gap. To this end, several methods have been recently proposed to automatically tune simulator parameters with posterior distributions given real data, for use with domain randomization at training time. These approaches have been shown to work for various robotic tasks under different settings and assumptions. Nevertheless, existing literature lacks a thorough comparison of existing adaptive domain randomization methods with respect to transfer performance and real-data efficiency. In this work, we present an open benchmark for both offline and online methods (SimOpt, BayRn, DROID, DROPO), to shed light on which are most suitable for each setting and task at hand. We found that online methods are limited by the quality of the currently learned policy for the next iteration, while offline methods may sometimes fail when replaying trajectories in simulation with open-loop commands. The code used will be released at https://github.com/gabrieletiboni/adr-benchmark.

Gabriele Goletto, Mirco Planamente, Barbara Caputo et al.

To enable a safe and effective human-robot cooperation, it is crucial to develop models for the identification of human activities. Egocentric vision seems to be a viable solution to solve this problem, and therefore many works provide deep learning solutions to infer human actions from first person videos. However, although very promising, most of these do not consider the major challenges that comes with a realistic deployment, such as the portability of the model, the need for real-time inference, and the robustness with respect to the novel domains (i.e., new spaces, users, tasks). With this paper, we set the boundaries that egocentric vision models should consider for realistic applications, defining a novel setting of egocentric action recognition in the wild, which encourages researchers to develop novel, applications-aware solutions. We also present a new model-agnostic technique that enables the rapid repurposing of existing architectures in this new context, demonstrating the feasibility to deploy a model on a tiny device (Jetson Nano) and to perform the task directly on the edge with very low energy consumption (2.4W on average at 50 fps). The code is publicly available at: https://github.com/EgocentricVision/EgoWild.

Gabriele Goletto, Tushar Nagarajan, Giuseppe Averta et al.

Egocentric videos provide a unique perspective into individuals' daily experiences, yet their unstructured nature presents challenges for perception. In this paper, we introduce AMEGO, a novel approach aimed at enhancing the comprehension of very-long egocentric videos. Inspired by the human's ability to maintain information from a single watching, AMEGO focuses on constructing a self-contained representations from one egocentric video, capturing key locations and object interactions. This representation is semantic-free and facilitates multiple queries without the need to reprocess the entire visual content. Additionally, to evaluate our understanding of very-long egocentric videos, we introduce the new Active Memories Benchmark (AMB), composed of more than 20K of highly challenging visual queries from EPIC-KITCHENS. These queries cover different levels of video reasoning (sequencing, concurrency and temporal grounding) to assess detailed video understanding capabilities. We showcase improved performance of AMEGO on AMB, surpassing other video QA baselines by a substantial margin.

Gabriele Tiboni, Andrea Protopapa, Tatiana Tommasi et al.

Soft robots are gaining popularity thanks to their intrinsic safety to contacts and adaptability. However, the potentially infinite number of Degrees of Freedom makes their modeling a daunting task, and in many cases only an approximated description is available. This challenge makes reinforcement learning (RL) based approaches inefficient when deployed on a realistic scenario, due to the large domain gap between models and the real platform. In this work, we demonstrate, for the first time, how Domain Randomization (DR) can solve this problem by enhancing RL policies for soft robots with: i) robustness w.r.t. unknown dynamics parameters; ii) reduced training times by exploiting drastically simpler dynamic models for learning; iii) better environment exploration, which can lead to exploitation of environmental constraints for optimal performance. Moreover, we introduce a novel algorithmic extension to previous adaptive domain randomization methods for the automatic inference of dynamics parameters for deformable objects. We provide an extensive evaluation in simulation on four different tasks and two soft robot designs, opening interesting perspectives for future research on Reinforcement Learning for closed-loop soft robot control.

Niccolò Cavagnero, Fernando Dos Santos, Marco Ciccone et al.

Deep Neural Networks (DNNs) enable a wide series of technological advancements, ranging from clinical imaging, to predictive industrial maintenance and autonomous driving. However, recent findings indicate that transient hardware faults may corrupt the models prediction dramatically. For instance, the radiation-induced misprediction probability can be so high to impede a safe deployment of DNNs models at scale, urging the need for efficient and effective hardening solutions. In this work, we propose to tackle the reliability issue both at training and model design time. First, we show that vanilla models are highly affected by transient faults, that can induce a performances drop up to 37%. Hence, we provide three zero-overhead solutions, based on DNN re-design and re-train, that can improve DNNs reliability to transient faults up to one order of magnitude. We complement our work with extensive ablation studies to quantify the gain in performances of each hardening component.

Simone Alberto Peirone, Gabriele Goletto, Mirco Planamente et al.

Human activities exhibit a strong correlation between actions and the places where these are performed, such as washing something at a sink. More specifically, in daily living environments we may identify particular locations, hereinafter named activity-centric zones, which may afford a set of homogeneous actions. Their knowledge can serve as a prior to favor vision models to recognize human activities. However, the appearance of these zones is scene-specific, limiting the transferability of this prior information to unfamiliar areas and domains. This problem is particularly relevant in egocentric vision, where the environment takes up most of the image, making it even more difficult to separate the action from the context. In this paper, we discuss the importance of decoupling the domain-specific appearance of activity-centric zones from their universal, domain-agnostic representations, and show how the latter can improve the cross-domain transferability of Egocentric Action Recognition (EAR) models. We validate our solution on the EPIC-Kitchens-100 and Argo1M datasets

Leonardo Iurada, Niccolò Cavagnero, Fernando Fernandes Dos Santos et al.

Deep learning models are crucial for autonomous vehicle perception, but their reliability is challenged by algorithmic limitations and hardware faults. We address the latter by examining fault-tolerance in semantic segmentation models. Using established hardware fault models, we evaluate existing hardening techniques both in terms of accuracy and uncertainty and introduce ReLUMax, a novel simple activation function designed to enhance resilience against transient faults. ReLUMax integrates seamlessly into existing architectures without time overhead. Our experiments demonstrate that ReLUMax effectively improves robustness, preserving performance and boosting prediction confidence, thus contributing to the development of reliable autonomous driving systems.

Francesca Pistilli, Giuseppe Averta

Deep neural networks for graphs have emerged as a powerful tool for learning on complex non-euclidean data, which is becoming increasingly common for a variety of different applications. Yet, although their potential has been widely recognised in the machine learning community, graph learning is largely unexplored for downstream tasks such as robotics applications. To fully unlock their potential, hence, we propose a review of graph neural architectures from a robotics perspective. The paper covers the fundamentals of graph-based models, including their architecture, training procedures, and applications. It also discusses recent advancements and challenges that arise in applied settings, related for example to the integration of perception, decision-making, and control. Finally, the paper provides an extensive review of various robotic applications that benefit from learning on graph structures, such as bodies and contacts modelling, robotic manipulation, action recognition, fleet motion planning, and many more. This survey aims to provide readers with a thorough understanding of the capabilities and limitations of graph neural architectures in robotics, and to highlight potential avenues for future research.

Claudia Cuttano, Gabriele Trivigno, Gabriele Rosi et al.

Referring Video Object Segmentation (RVOS) relies on natural language expressions to segment an object in a video clip. Existing methods restrict reasoning either to independent short clips, losing global context, or process the entire video offline, impairing their application in a streaming fashion. In this work, we aim to surpass these limitations and design an RVOS method capable of effectively operating in streaming-like scenarios while retaining contextual information from past frames. We build upon the Segment-Anything 2 (SAM2) model, that provides robust segmentation and tracking capabilities and is naturally suited for streaming processing. We make SAM2 wiser, by empowering it with natural language understanding and explicit temporal modeling at the feature extraction stage, without fine-tuning its weights, and without outsourcing modality interaction to external models. To this end, we introduce a novel adapter module that injects temporal information and multi-modal cues in the feature extraction process. We further reveal the phenomenon of tracking bias in SAM2 and propose a learnable module to adjust its tracking focus when the current frame features suggest a new object more aligned with the caption. Our proposed method, SAMWISE, achieves state-of-the-art across various benchmarks, by adding a negligible overhead of less than 5 M parameters. Code is available at https://github.com/ClaudiaCuttano/SAMWISE .

Giorgia Modi, Davide Buoso, Giuseppe Averta et al.

Current approaches to 3D scene graph generation rely on dedicated depth sensors, such as LiDAR or RGB-D cameras, for metric 3D reconstruction. This limits deployment to specialized robotic platforms and excludes settings where only RGB cameras are available, such as fixed external infrastructure. Existing pipelines also typically operate on passively collected observation trajectories, rather than selecting viewpoints based on the partially built scene representation, and therefore fail to effectively exploit the semantic and spatial information encoded within the graph during exploration. This paper presents a fully visual framework for the active, incremental construction of 3D scene graphs from RGB input only, addressing both limitations. The proposed approach unifies perception and planning around a shared structured representation that captures object semantics, 3D geometry, relational context, and information from multiple viewpoints. Because the framework is hardware-agnostic and relies only on RGB observations, it can incorporate inputs from both onboard robot cameras and fixed external cameras within the same representation. Experiments on the Replica dataset show that the RGB-only pipeline achieves F1-score parity with baselines using ground-truth depth. Active exploration experiments on ReplicaCAD further show that semantic-driven viewpoint selection detects more than twice as many objects as a geometric frontier-based baseline under the same exploration budget. Finally, the external-camera setting demonstrates that complementary RGB views can effectively bootstrap the scene graph and improve contextual understanding at no additional exploration cost.

Giorgia Modi, Davide Buoso, Giuseppe Averta et al.

Commonly available prior information, such as BIM models, floor plans, and remote sensing images, can provide valuable geometric and semantic context for autonomous robotic systems. In this paper, we treat observations from fixed external RGB cameras as Common Prior Maps (CPMs): wide-field views of the environment that initialize a semantic and geometric scene prior before any robot motion begins. We present an RGB-only framework for active, incremental 3D scene graph (3DSG) generation that seamlessly fuses observations from both onboard robot cameras and fixed external cameras within a single hardware-agnostic pipeline. By relying solely on RGB observations processed by a feed-forward 3D reconstruction model, the system treats all cameras - onboard or external - identically, requiring no hardware modifications. A graph-based active semantic exploration framework then directly leverages the partial scene graph to guide the robot toward regions of high semantic uncertainty, progressively completing and refining the prior. Experiments demonstrate that bootstrapping the scene graph with even a single external camera increases initial object recall by up to +79%, and that the richer context of the prior significantly improves the efficiency of subsequent active exploration.

Niccolò Cavagnero, Luca Robbiano, Francesca Pistilli et al.

Neural Networks design is a complex and often daunting task, particularly for resource-constrained scenarios typical of mobile-sized models. Neural Architecture Search is a promising approach to automate this process, but existing competitive methods require large training time and computational resources to generate accurate models. To overcome these limits, this paper contributes with: i) a novel training-free metric, named Entropic Score, to estimate model expressivity through the aggregated element-wise entropy of its activations; ii) a cyclic search algorithm to separately yet synergistically search model size and topology. Entropic Score shows remarkable ability in searching for the topology of the network, and a proper combination with LogSynflow, to search for model size, yields superior capability to completely design high-performance Hybrid Transformers for edge applications in less than 1 GPU hour, resulting in the fastest and most accurate NAS method for ImageNet classification.

Mirco Planamente, Gabriele Goletto, Gabriele Trivigno et al.

In this report, we describe the technical details of our submission to the EPIC-Kitchens-100 Unsupervised Domain Adaptation (UDA) Challenge in Action Recognition. To tackle the domain-shift which exists under the UDA setting, we first exploited a recent Domain Generalization (DG) technique, called Relative Norm Alignment (RNA). Secondly, we extended this approach to work on unlabelled target data, enabling a simpler adaptation of the model to the target distribution in an unsupervised fashion. To this purpose, we included in our framework UDA algorithms, such as multi-level adversarial alignment and attentive entropy. By analyzing the challenge setting, we notice the presence of a secondary concurrence shift in the data, which is usually called environmental bias. It is caused by the existence of different environments, i.e., kitchens. To deal with these two shifts (environmental and temporal), we extended our system to perform Multi-Source Multi-Target Domain Adaptation. Finally, we employed distinct models in our final proposal to leverage the potential of popular video architectures, and we introduced two more losses for the ensemble adaptation. Our submission (entry 'plnet') is visible on the leaderboard and ranked in 2nd position for 'verb', and in 3rd position for both 'noun' and 'action'.

Davide Buoso, Andrea Protopapa, Stefano Di Carlo et al.

Learning visuomotor policies from scarce expert demonstrations remains a core challenge in robotic manipulation. A primary hurdle lies in distilling high-dimensional RGB representations into control-relevant geometry without overfitting. While using frozen pre-trained Vision Foundation Models (VFMs) improves data efficiency, it also shifts most task adaptation onto a small spatial pooling module, which can latch onto task-irrelevant shortcuts and lose geometric grounding when finetuned with few data samples. More broadly, pre-trained visual representations used for policy learning have been observed to struggle under even minor scene perturbations, highlighting the need for robustness-oriented inductive biases. We propose Geometric Anchor Pre-training (GAP), a simple, action-free warm-up stage that regularizes the spatial adapter before downstream imitation learning. GAP pre-trains the pooling layer on a lightweight simulated proxy task where object masks are available at no cost, encouraging the adapter to produce keypoints that lie on the object, cover its spatial extent, and remain sharp and repeatable over time. This yields stable geometric anchors that provide a reliable coordinate interface for few-shot policy learning, while keeping the VFM frozen. We evaluate GAP on RoboMimic and ManiSkill under severe data scarcity (15-50 demonstrations) and domain shift. A simple adapter regularized with GAP consistently outperforms stronger attention-based poolers and end-to-end fine-tuning, achieving 62% success on RoboMimic Can with 15 demonstrations (+16% over AFA), 63% on the long-horizon high-precision Tool Hang task with 50 demonstrations, and 61% on ManiSkill StackCube with 30 demonstrations (+11% over full fine-tuning). The proxy stage is lightweight and fully decoupled from downstream tasks, making it practical to reuse across environments and manipulation skills.

Claudia Cuttano, Gabriele Trivigno, Giuseppe Averta et al.

Few-shot segmentation aims to segment unseen object categories from just a handful of annotated examples. This requires mechanisms that can both identify semantically related objects across images and accurately produce segmentation masks. We note that Segment Anything 2 (SAM2), with its prompt-and-propagate mechanism, offers both strong segmentation capabilities and a built-in feature matching process. However, we show that its representations are entangled with task-specific cues optimized for object tracking, which impairs its use for tasks requiring higher level semantic understanding. Our key insight is that, despite its class-agnostic pretraining, SAM2 already encodes rich semantic structure in its features. We propose SANSA (Semantically AligNed Segment Anything 2), a framework that makes this latent structure explicit, and repurposes SAM2 for few-shot segmentation through minimal task-specific modifications. SANSA achieves state-of-the-art performance on few-shot segmentation benchmarks specifically designed to assess generalization, outperforms generalist methods in the popular in-context setting, supports various prompts flexible interaction via points, boxes, or scribbles, and remains significantly faster and more compact than prior approaches. Code is available at https://github.com/ClaudiaCuttano/SANSA.

Niccolò Cavagnero, Gabriele Rosi, Claudia Cuttano et al.

Recent transformer-based architectures have shown impressive results in the field of image segmentation. Thanks to their flexibility, they obtain outstanding performance in multiple segmentation tasks, such as semantic and panoptic, under a single unified framework. To achieve such impressive performance, these architectures employ intensive operations and require substantial computational resources, which are often not available, especially on edge devices. To fill this gap, we propose Prototype-based Efficient MaskFormer (PEM), an efficient transformer-based architecture that can operate in multiple segmentation tasks. PEM proposes a novel prototype-based cross-attention which leverages the redundancy of visual features to restrict the computation and improve the efficiency without harming the performance. In addition, PEM introduces an efficient multi-scale feature pyramid network, capable of extracting features that have high semantic content in an efficient way, thanks to the combination of deformable convolutions and context-based self-modulation. We benchmark the proposed PEM architecture on two tasks, semantic and panoptic segmentation, evaluated on two different datasets, Cityscapes and ADE20K. PEM demonstrates outstanding performance on every task and dataset, outperforming task-specific architectures while being comparable and even better than computationally-expensive baselines.

Tommie Kerssies, Niccolò Cavagnero, Alexander Hermans et al.

Vision Transformers (ViTs) have shown remarkable performance and scalability across various computer vision tasks. To apply single-scale ViTs to image segmentation, existing methods adopt a convolutional adapter to generate multi-scale features, a pixel decoder to fuse these features, and a Transformer decoder that uses the fused features to make predictions. In this paper, we show that the inductive biases introduced by these task-specific components can instead be learned by the ViT itself, given sufficiently large models and extensive pre-training. Based on these findings, we introduce the Encoder-only Mask Transformer (EoMT), which repurposes the plain ViT architecture to conduct image segmentation. With large-scale models and pre-training, EoMT obtains a segmentation accuracy similar to state-of-the-art models that use task-specific components. At the same time, EoMT is significantly faster than these methods due to its architectural simplicity, e.g., up to 4x faster with ViT-L. Across a range of model sizes, EoMT demonstrates an optimal balance between segmentation accuracy and prediction speed, suggesting that compute resources are better spent on scaling the ViT itself rather than adding architectural complexity. Code: https://www.tue-mps.org/eomt/.

Claudia Cuttano, Gabriele Rosi, Gabriele Trivigno et al.

Humans show an innate capability to identify tools to support specific actions. The association between objects parts and the actions they facilitate is usually named affordance. Being able to segment objects parts depending on the tasks they afford is crucial to enable intelligent robots to use objects of daily living. Traditional supervised learning methods for affordance segmentation require costly pixel-level annotations, while weakly supervised approaches, though less demanding, still rely on object-interaction examples and support a closed set of actions. These limitations hinder scalability, may introduce biases, and usually restrict models to a limited set of predefined actions. This paper proposes AffordanceCLIP, to overcome these limitations by leveraging the implicit affordance knowledge embedded within large pre-trained Vision-Language models like CLIP. We experimentally demonstrate that CLIP, although not explicitly trained for affordances detection, retains valuable information for the task. Our AffordanceCLIP achieves competitive zero-shot performance compared to methods with specialized training, while offering several advantages: i) it works with any action prompt, not just a predefined set; ii) it requires training only a small number of additional parameters compared to existing solutions and iii) eliminates the need for direct supervision on action-object pairs, opening new perspectives for functionality-based reasoning of models.

Simone Alberto Peirone, Francesca Pistilli, Antonio Alliegro et al.

Human comprehension of a video stream is naturally broad: in a few instants, we are able to understand what is happening, the relevance and relationship of objects, and forecast what will follow in the near future, everything all at once. We believe that - to effectively transfer such an holistic perception to intelligent machines - an important role is played by learning to correlate concepts and to abstract knowledge coming from different tasks, to synergistically exploit them when learning novel skills. To accomplish this, we seek for a unified approach to video understanding which combines shared temporal modelling of human actions with minimal overhead, to support multiple downstream tasks and enable cooperation when learning novel skills. We then propose EgoPack, a solution that creates a collection of task perspectives that can be carried across downstream tasks and used as a potential source of additional insights, as a backpack of skills that a robot can carry around and use when needed. We demonstrate the effectiveness and efficiency of our approach on four Ego4D benchmarks, outperforming current state-of-the-art methods.

Gabriele Rosi, Claudia Cuttano, Niccolò Cavagnero et al.

Recent advancements in image segmentation have focused on enhancing the efficiency of the models to meet the demands of real-time applications, especially on edge devices. However, existing research has primarily concentrated on single-task settings, especially on semantic segmentation, leading to redundant efforts and specialized architectures for different tasks. To address this limitation, we propose a novel architecture for efficient multi-task image segmentation, capable of handling various segmentation tasks without sacrificing efficiency or accuracy. We introduce BiSeNetFormer, that leverages the efficiency of two-stream semantic segmentation architectures and it extends them into a mask classification framework. Our approach maintains the efficient spatial and context paths to capture detailed and semantic information, respectively, while leveraging an efficient transformed-based segmentation head that computes the binary masks and class probabilities. By seamlessly supporting multiple tasks, namely semantic and panoptic segmentation, BiSeNetFormer offers a versatile solution for multi-task segmentation. We evaluate our approach on popular datasets, Cityscapes and ADE20K, demonstrating impressive inference speeds while maintaining competitive accuracy compared to state-of-the-art architectures. Our results indicate that BiSeNetFormer represents a significant advancement towards fast, efficient, and multi-task segmentation networks, bridging the gap between model efficiency and task adaptability.

Simone Alberto Peirone, Francesca Pistilli, Antonio Alliegro et al.

Our comprehension of video streams depicting human activities is naturally multifaceted: in just a few moments, we can grasp what is happening, identify the relevance and interactions of objects in the scene, and forecast what will happen soon, everything all at once. To endow autonomous systems with such a holistic perception, learning how to correlate concepts, abstract knowledge across diverse tasks, and leverage tasks synergies when learning novel skills is essential. A significant step in this direction is EgoPack, a unified framework for understanding human activities across diverse tasks with minimal overhead. EgoPack promotes information sharing and collaboration among downstream tasks, essential for efficiently learning new skills. In this paper, we introduce Hier-EgoPack, which advances EgoPack by enabling reasoning also across diverse temporal granularities, which expands its applicability to a broader range of downstream tasks. To achieve this, we propose a novel hierarchical architecture for temporal reasoning equipped with a GNN layer specifically designed to tackle the challenges of multi-granularity reasoning effectively. We evaluate our approach on multiple Ego4d benchmarks involving both clip-level and frame-level reasoning, demonstrating how our hierarchical unified architecture effectively solves these diverse tasks simultaneously.

Davide Buoso, Luke Robinson, Giuseppe Averta et al.

This study explores the potential of off-the-shelf Vision-Language Models (VLMs) for high-level robot planning in the context of autonomous navigation. Indeed, while most of existing learning-based approaches for path planning require extensive task-specific training/fine-tuning, we demonstrate how such training can be avoided for most practical cases. To do this, we introduce Select2Plan (S2P), a novel training-free framework for high-level robot planning which completely eliminates the need for fine-tuning or specialised training. By leveraging structured Visual Question-Answering (VQA) and In-Context Learning (ICL), our approach drastically reduces the need for data collection, requiring a fraction of the task-specific data typically used by trained models, or even relying only on online data. Our method facilitates the effective use of a generally trained VLM in a flexible and cost-efficient way, and does not require additional sensing except for a simple monocular camera. We demonstrate its adaptability across various scene types, context sources, and sensing setups. We evaluate our approach in two distinct scenarios: traditional First-Person View (FPV) and infrastructure-driven Third-Person View (TPV) navigation, demonstrating the flexibility and simplicity of our method. Our technique significantly enhances the navigational capabilities of a baseline VLM of approximately 50% in TPV scenario, and is comparable to trained models in the FPV one, with as few as 20 demonstrations.

Claudia Cuttano, Antonio Tavera, Fabio Cermelli et al.

Many practical applications require training of semantic segmentation models on unlabelled datasets and their execution on low-resource hardware. Distillation from a trained source model may represent a solution for the first but does not account for the different distribution of the training data. Unsupervised domain adaptation (UDA) techniques claim to solve the domain shift, but in most cases assume the availability of the source data or an accessible white-box source model, which in practical applications are often unavailable for commercial and/or safety reasons. In this paper, we investigate a more challenging setting in which a lightweight model has to be trained on a target unlabelled dataset for semantic segmentation, under the assumption that we have access only to black-box source model predictions. Our method, named CoRTe, consists of (i) a pseudo-labelling function that extracts reliable knowledge from the black-box source model using its relative confidence, (ii) a pseudo label refinement method to retain and enhance the novel information learned by the student model on the target data, and (iii) a consistent training of the model using the extracted pseudo labels. We benchmark CoRTe on two synthetic-to-real settings, demonstrating remarkable results when using black-box models to transfer knowledge on lightweight models for a target data distribution.

Amirshayan Nasirimajd, Chiara Plizzari, Simone Alberto Peirone et al.

Recognizing human activities from visual inputs, particularly through a first-person viewpoint, is essential for enabling robots to replicate human behavior. Egocentric vision, characterized by cameras worn by observers, captures diverse changes in illumination, viewpoint, and environment. This variability leads to a notable drop in the performance of Egocentric Action Recognition models when tested in environments not seen during training. In this paper, we tackle these challenges by proposing a domain generalization approach for Egocentric Action Recognition. Our insight is that action sequences often reflect consistent user intent across visual domains. By leveraging action sequences, we aim to enhance the model's generalization ability across unseen environments. Our proposed method, named SeqDG, introduces a visual-text sequence reconstruction objective (SeqRec) that uses contextual cues from both text and visual inputs to reconstruct the central action of the sequence. Additionally, we enhance the model's robustness by training it on mixed sequences of actions from different domains (SeqMix). We validate SeqDG on the EGTEA and EPIC-KITCHENS-100 datasets. Results on EPIC-KITCHENS-100, show that SeqDG leads to +2.4% relative average improvement in cross-domain action recognition in unseen environments, and on EGTEA the model achieved +0.6% Top-1 accuracy over SOTA in intra-domain action recognition.

Simone Alberto Peirone, Francesca Pistilli, Antonio Alliegro et al.

Our comprehension of video streams depicting human activities is naturally multifaceted: in just a few moments, we can grasp what is happening, identify the relevance and interactions of objects in the scene, and forecast what will happen soon, everything all at once. To endow autonomous systems with such holistic perception, learning how to correlate concepts, abstract knowledge across diverse tasks, and leverage tasks synergies when learning novel skills is essential. In this paper, we introduce Hier-EgoPack, a unified framework able to create a collection of task perspectives that can be carried across downstream tasks and used as a potential source of additional insights, as a backpack of skills that a robot can carry around and use when needed.

Simone Alberto Peirone, Francesca Pistilli, Giuseppe Averta

Human activities are particularly complex and variable, and this makes challenging for deep learning models to reason about them. However, we note that such variability does have an underlying structure, composed of a hierarchy of patterns of related actions. We argue that such structure can emerge naturally from unscripted videos of human activities, and can be leveraged to better reason about their content. We present HiERO, a weakly-supervised method to enrich video segments features with the corresponding hierarchical activity threads. By aligning video clips with their narrated descriptions, HiERO infers contextual, semantic and temporal reasoning with an hierarchical architecture. We prove the potential of our enriched features with multiple video-text alignment benchmarks (EgoMCQ, EgoNLQ) with minimal additional training, and in zero-shot for procedure learning tasks (EgoProceL and Ego4D Goal-Step). Notably, HiERO achieves state-of-the-art performance in all the benchmarks, and for procedure learning tasks it outperforms fully-supervised methods by a large margin (+12.5% F1 on EgoProceL) in zero shot. Our results prove the relevance of using knowledge of the hierarchy of human activities for multiple reasoning tasks in egocentric vision.

Francesco Capuano, Gabriele Tiboni, Niccolò Cavagnero et al.

Existing hardware-aware NAS (HW-NAS) methods typically assume access to precise information circa the target device, either via analytical approximations of the post-compilation latency model, or through learned latency predictors. Such approximate approaches risk introducing estimation errors that may prove detrimental in risk-sensitive applications. In this work, we propose a two-stage HW-NAS framework, in which we first learn an architecture controller on a distribution of synthetic devices, and then directly deploy the controller on a target device. At test-time, our network controller deploys directly to the target device without relying on any pre-collected information, and only exploits direct interactions. In particular, the pre-training phase on synthetic devices enables the controller to design an architecture for the target device by interacting with it through a small number of high-fidelity latency measurements. To guarantee accessibility of our method, we only train our controller with training-free accuracy proxies, allowing us to scale the meta-training phase without incurring the overhead of full network training. We benchmark on HW-NATS-Bench, demonstrating that our method generalizes to unseen devices and searches for latency-efficient architectures by in-context adaptation using only a few real-world latency evaluations at test-time.

Antonio Alliegro, Francesca Pistilli, Tatiana Tommasi et al.

Forecasting human-environment interactions in daily activities is challenging due to the high variability of human behavior. While predicting directly from videos is possible, it is limited by confounding factors like irrelevant objects or background noise that do not contribute to the interaction. A promising alternative is using Scene Graphs (SGs) to track only the relevant elements. However, current methods for forecasting future SGs face significant challenges and often rely on unrealistic assumptions, such as fixed objects over time, limiting their applicability to long-term activities where interacted objects may appear or disappear. In this paper, we introduce FORESCENE, a novel framework for Scene Graph Anticipation (SGA) that predicts both object and relationship evolution over time. FORESCENE encodes observed video segments into a latent representation using a tailored Graph Auto-Encoder and forecasts future SGs using a Latent Diffusion Model (LDM). Our approach enables continuous prediction of interaction dynamics without making assumptions on the graph's content or structure. We evaluate FORESCENE on the Action Genome dataset, where it outperforms existing SGA methods while solving a significantly more complex task.