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.

Astronauts take thousands of photos of Earth per day from the International Space Station, which, once localized on Earth's surface, are used for a multitude of tasks, ranging from climate change research to disaster management. The localization process, which has been performed manually for decades, has recently been approached through image retrieval solutions: given an astronaut photo, find its most similar match among a large database of geo-tagged satellite images, in a task called Astronaut Photography Localization (APL). Yet, existing APL approaches are trained only using satellite images, without taking advantage of the millions open-source astronaut photos. In this work we present the first APL pipeline capable of leveraging astronaut photos for training. We first produce full localization information for 300,000 manually weakly labeled astronaut photos through an automated pipeline, and then use these images to train a model, called AstroLoc. AstroLoc learns a robust representation of Earth's surface features through two losses: astronaut photos paired with their matching satellite counterparts in a pairwise loss, and a second loss on clusters of satellite imagery weighted by their relevance to astronaut photography via unsupervised mining. We find that AstroLoc achieves a staggering 35% average improvement in recall@1 over previous SOTA, pushing the limits of existing datasets with a recall@100 consistently over 99%. Finally, we note that AstroLoc, without any fine-tuning, provides excellent results for related tasks like the lost-in-space satellite problem and historical space imagery localization.

Autonomous driving (AD) systems are becoming increasingly capable of handling complex tasks, mainly due to recent advances in deep learning and AI. As interactions between autonomous systems and humans increase, the interpretability of decision-making processes in driving systems becomes increasingly crucial for ensuring safe driving operations. Successful human-machine interaction requires understanding the underlying representations of the environment and the driving task, which remains a significant challenge in deep learning-based systems. To address this, we introduce the task of interpretability in maneuver prediction before they occur for driver safety, i.e., driver intent prediction (DIP), which plays a critical role in AD systems. To foster research in interpretable DIP, we curate the eXplainable Driving Action Anticipation Dataset (DAAD-X), a new multimodal, ego-centric video dataset to provide hierarchical, high-level textual explanations as causal reasoning for the driver's decisions. These explanations are derived from both the driver's eye-gaze and the ego-vehicle's perspective. Next, we propose Video Concept Bottleneck Model (VCBM), a framework that generates spatio-temporally coherent explanations inherently, without relying on post-hoc techniques. Finally, through extensive evaluations of the proposed VCBM on the DAAD-X dataset, we demonstrate that transformer-based models exhibit greater interpretability than conventional CNN-based models. Additionally, we introduce a multilabel t-SNE visualization technique to illustrate the disentanglement and causal correlation among multiple explanations. Our data, code and models are available at: https://mukil07.github.io/VCBM.github.io/

Few-shot segmentation aims to segment unseen categories from just a handful of annotated examples. This requires mechanisms to identify semantically related objects across images and accurately produce masks. We note that Segment Anything 2 (SAM2), with its prompt-and-propagate mechanism, provides 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 on few-shot segmentation benchmarks designed to assess generalization and outperforms generalist methods in the popular in-context setting. Additionally, it supports flexible promptable interaction via points, boxes, or scribbles, and remains significantly faster and more compact than prior approaches. Code at: https://github.com/ClaudiaCuttano/SANSA.

Astronaut photography, spanning six decades of human spaceflight, presents a unique Earth observations dataset with immense value for both scientific research and disaster response. Despite their significance, accurately localizing the geographical extent of these images, which is crucial for effective utilization, poses substantial challenges. Cur-rent, manual localization efforts are time-consuming, mo-tivating the need for automated solutions. We propose a novel approach - leveraging image retrieval - to address this challenge efficiently. We introduce innovative training techniques which contribute to the development of a high-performance model, EarthLoc. We develop six evaluation datasets and perform a comprehensive benchmark comparing EarthLoc to existing methods, showcasing its superior efficiency and accuracy. Our approach marks a signifi-cant advancement in automating the localization of astro-naut photography, which will help bridge a critical gap in Earth observations data. Code and datasets are available at https://github.com/gmberton/EarthLoc.

Pose refinement is an interesting and practically rele-vant research direction. Pose refinement can be used to (1) obtain a more accurate pose estimate from an initial prior (e.g., from retrieval), (2) as pre-processing, i.e., to provide a better starting point to a more expensive pose estimator, (3) as post-processing of a more accurate local-izer. Existing approaches focus on learning features / scene representations for the pose refinement task. This involves training an implicit scene representation or learning features while optimizing a camera pose-based loss. A natural question is whether training specific features / representations is truly necessary or whether similar results can be al-ready achieved with more generic features. In this work, we present a simple approach that combines pre-trained features with a particle filter and a renderable representation of the scene. Despite its simplicity, it achieves state-of-the-art results, demonstrating that one can easily build a pose refiner without the need for specific training. The code is at https://github.com/gali130/mcloc_poseref

In computer vision, keypoint detection is a fundamental task, with applications spanning from robotics to image retrieval; however, existing learning-based methods suffer from scale dependency, and lack flexibility. This paper introduces a novel approach that leverages Morse theory and persistent homology, powerful tools rooted in algebraic topology. We propose a novel loss function based on the recent introduction of a notion of subgradient in persistent homology, paving the way towards topological learning. Our detector, MorseDet, is the first topology-based learning model for feature detection, which achieves competitive performance in keypoint repeatability and introduces a principled and theoretically robust approach to the problem.

Segmenting unknown or anomalous object instances is a critical task in autonomous driving applications, and it is approached traditionally as a per-pixel classification problem. However, reasoning individually about each pixel without considering their contextual semantics results in high uncertainty around the objects' boundaries and numerous false positives. We propose a paradigm change by shifting from a per-pixel classification to a mask classification. Our mask-based method, Mask2Anomaly, demonstrates the feasibility of integrating a mask-classification architecture to jointly address anomaly segmentation, open-set semantic segmentation, and open-set panoptic segmentation. Mask2Anomaly includes several technical novelties that are designed to improve the detection of anomalies/unknown objects: i) a global masked attention module to focus individually on the foreground and background regions; ii) a mask contrastive learning that maximizes the margin between an anomaly and known classes; iii) a mask refinement solution to reduce false positives; and iv) a novel approach to mine unknown instances based on the mask- architecture properties. By comprehensive qualitative and qualitative evaluation, we show Mask2Anomaly achieves new state-of-the-art results across the benchmarks of anomaly segmentation, open-set semantic segmentation, and open-set panoptic segmentation.

Visual Place Recognition is a task that aims to predict the place of an image (called query) based solely on its visual features. This is typically done through image retrieval, where the query is matched to the most similar images from a large database of geotagged photos, using learned global descriptors. A major challenge in this task is recognizing places seen from different viewpoints. To overcome this limitation, we propose a new method, called EigenPlaces, to train our neural network on images from different point of views, which embeds viewpoint robustness into the learned global descriptors. The underlying idea is to cluster the training data so as to explicitly present the model with different views of the same points of interest. The selection of this points of interest is done without the need for extra supervision. We then present experiments on the most comprehensive set of datasets in literature, finding that EigenPlaces is able to outperform previous state of the art on the majority of datasets, while requiring 60% less GPU memory for training and using 50% smaller descriptors. The code and trained models for EigenPlaces are available at https://github.com/gmberton/EigenPlaces, while results with any other baseline can be computed with the codebase at https://github.com/gmberton/auto_VPR.

Anomaly segmentation is a critical task for driving applications, and it is approached traditionally as a per-pixel classification problem. However, reasoning individually about each pixel without considering their contextual semantics results in high uncertainty around the objects’ boundaries and numerous false positives. We propose a paradigm change by shifting from a per-pixel classification to a mask classification. Our mask-based method, Mask2Anomaly, demonstrates the feasibility of integrating an anomaly detection method in a mask-classification architecture. Mask2Anomaly includes several technical novelties that are designed to improve the detection of anomalies in masks: i) a global masked attention module to focus individually on the foreground and background regions; ii) a mask contrastive learning that maximizes the margin between an anomaly and known classes; and iii) a mask refinement solution to reduce false positives. Mask2Anomaly achieves new state-of-the-art results across a range of benchmarks, both in the per-pixel and component-level evaluations. In particular, Mask2Anomaly reduces the average false positives rate by 60% w.r.t. the previous state-of-the-art.

Visual Place recognition is commonly addressed as an image retrieval problem. However, retrieval methods are impractical to scale to large datasets, densely sampled from city-wide maps, since their dimension impact negatively on the inference time. Using approximate nearest neighbour search for retrieval helps to mitigate this issue, at the cost of a performance drop. In this paper we investigate whether we can effectively approach this task as a classification problem, thus bypassing the need for a similarity search. We find that existing classification methods for coarse, planet-wide localization are not suitable for the fine-grained and city-wide setting. This is largely due to how the dataset is split into classes, because these methods are designed to handle a sparse distribution of photos and as such do not consider the visual aliasing problem across neighbouring classes that naturally arises in dense scenarios. Thus, we propose a partitioning scheme that enables a fast and accurate inference, preserving a simple learning procedure, and a novel inference pipeline based on an ensemble of novel classifiers that uses the prototypes learned via an angular margin loss. Our method, Divide&Classify (D&C), enjoys the fast inference of classification solutions and an accuracy competitive with retrieval methods on the fine-grained, city-wide setting. Moreover, we show that D&C can be paired with existing retrieval pipelines to speed up computations by over 20 times while increasing their recall, leading to new state-of-the-art results. Code is available at https://github.com/ga1i13o/Divide-and-Classify

Visual Geo-localization (VG) is the task of estimating the position where a given photo was taken by comparing it with a large database of images of known locations. To investigate how existing techniques would perform on a real-world city-wide VG application, we build San Francisco eXtra Large, a new dataset covering a whole city and providing a wide range of challenging cases, with a size 30x bigger than the previous largest dataset for visual geo-localization. We find that current methods fail to scale to such large datasets, therefore we design a new highly scalable training technique, called CosPlace, which casts the training as a classification problem avoiding the expensive mining needed by the commonly used contrastive learning. We achieve state-of-the-art performance on a wide range of datasets and find that CosPlace is robust to heavy domain changes. Moreover, we show that, compared to the previous state-of-the-art, CosPlace requires roughly 80% less GPU memory at train time, and it achieves better results with 8x smaller descriptors, paving the way for city-wide real-world visual geo-localization. Dataset, code and trained models are available for research purposes at https://github.com/gmberton/CosPlace.

In this paper, we propose a new open-source benchmarkingframeworkfor Visual Geo-localization (VG) that allows to build, train, and test a wide range of commonly used ar-chitectures, with the flexibility to change individual components of a geo-localization pipeline. The purpose of this framework is twofold: i) gaining insights into how differ-ent components and design choices in a VG pipeline im-pact the final results, both in terms of performance (re-call@N metric) and system requirements (such as execution time and memory consumption); ii) establish a system-atic evaluation protocol for comparing different methods. Using the proposed framework, we perform a large suite of experiments which provide criteria for choosing back-bone, aggregation and negative mining depending on the use-case and requirements. We also assess the impact of engineering techniques like pre/post-processing, data aug-mentation and image resizing, showing that better performance can be obtained through somewhat simple procedures: for example, downscaling the images' resolution to 80% can lead to similar results with a 36% savings in ex-traction time and dataset storage requirement. Code and trained models are available at dataset storage requirement. https://deep-vg-bench.herokuapp.com/.

In this paper we consider the task of semantic segmentation in autonomous driving applications. Specifically, we consider the cross-domain few-shot setting where training can use only few real-world annotated images and many annotated synthetic images. In this context, aligning the domains is made more challenging by the pixel-wise class imbalance that is intrinsic in the segmentation and that leads to ignoring the underrepresented classes and overfitting the well represented ones. We address this problem with a novel framework called Pixel-By-Pixel Cross-Domain Alignment (PixDA). We propose a novel pixel-by-pixel domain adversarial loss following three criteria: (i) align the source and the target domain for each pixel, (ii) avoid negative transfer on the correctly represented pixels, and (iii) regularize the training of infrequent classes to avoid overfitting. The pixel-wise adversarial training is assisted by a novel sample selection procedure, that handles the imbalance between source and target data, and a knowledge distillation strategy, that avoids overfitting towards the few target images. We demonstrate on standard synthetic-to-real benchmarks that PixDA outperforms previous state-of-the-art methods in (1-5)-shot settings.1

In this paper we propose a novel method for image matching based on dense local features and tailored for visual geolocalization. Dense local features matching is robust against changes in illumination and occlusions, but not against viewpoint shifts which are a fundamental aspect of geolocalization. Our method, called GeoWarp, directly embeds invariance to viewpoint shifts in the process of extracting dense features. This is achieved via a trainable module which learns from the data an invariance that is meaningful for the task of recognizing places. We also devise a new self-supervised loss and two new weakly supervised losses to train this module using only unlabeled data and weak labels. GeoWarp is implemented efficiently as a re-ranking method that can be easily embedded into pre-existing visual geolocalization pipelines. Experimental validation on standard geolocalization benchmarks demonstrates that GeoWarp boosts the accuracy of state-of-the-art retrieval architectures. The code and trained models will be released upon acceptance of this paper.

We address the task of cross-domain visual place recognition, where the goal is to geolocalize a given query image against a labeled gallery, in the case where the query and the gallery belong to different visual domains. To achieve this, we focus on building a domain robust deep network by leveraging over an attention mechanism combined with few-shot unsupervised domain adaptation techniques, where we use a small number of unlabeled target domain images to learn about the target distribution. With our method, we are able to outperform the current state of the art while using two orders of magnitude less target domain images. Finally we propose a new large-scale dataset for cross-domain visual place recognition, called SVOX. The pytorch code is available at https://github.com/valeriopaolicelli/AdAGeo.

In this paper we present a robust quadrotor controller for tracking a reference trajectory in presence of uncertainties and disturbances. A Super Twisting controller is implemented using the recently proposed gain adaptation law [1], [2], which has the advantage of not requiring the knowledge of the upper bound of the lumped uncertainties. The controller design is based on the regular form of the quadrotor dynamics, without separation in two nested control loops for position and attitude. The controller is further extended by a feedforward dynamic inversion control that reduces the effort of the sliding mode controller. The higher order quadrotor dynamic model and proposed controller are validated using a SimMechanics physical simulation with initial error, parameter uncertainties, noisy measurements and external perturbations.

This paper addresses the problem of cooperative aerial transportation of an object using a team of quadrotors. The approach presented to solve this problem accounts for the full dynamics of the system and it is inspired by the literature on reconfigurable cable-driven parallel robots (RCDPR). Using the modelling convention of RCDPR it is derived a direct relation between the motion of the quadrotors and the motion of the payload. This relation makes explicit the available internal motion of the system, which can be used to automatically achieve additional tasks. The proposed method does not require to specify a priory the forces in the cables and uses a tension distribution algorithm to optimally distribute them among the robots. The presented framework is also suitable for online teleoperation. Physical simulations with a human-in-the-loop validate the proposed approach.

This paper introduces the CableRobot simulator, which was developed at the Max Planck Institute for Biological Cybernetics in cooperation with the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. The simulator is a completely novel approach to the design of motion simulation platforms in so far as it uses cables and winches for actuation instead of rigid links known from hexapod simulators. This approach allows to reduce the actuated mass, scale up the workspace significantly, and provides great flexibility to switch between system configurations in which the robot can be operated. The simulator will be used for studies in the field of human perception research and virtual reality applications. The paper discusses some of the issues arising from the usage of cables and provides a system overview regarding kinematics and system dynamics as well as giving a brief introduction into possible application use cases.

A new framework for semi-autonomous path planning for mobile robots that extends the classical paradigm of bilateral shared control is presented. The path is represented as a B-spline and the human operator can modify its shape by controlling the motion of a finite number of control points. An autonomous algorithm corrects in real time the human directives in order to facilitate path tracking for the mobile robot and ensures i) collision avoidance, ii) path regularity, and iii) attraction to nearby points of interest. A haptic feedback algorithm processes both human's and autonomous control terms, and their integrals, to provide an information of the mismatch between the path specified by the operator and the one corrected by the autonomous algorithm. The framework is validated with extensive experiments using a quadrotor UAV and a human in the loop with two haptic interfaces.

This work extends the framework of bilateral shared control of mobile robots with the aim of increasing the robot autonomy and decreasing the operator commitment. We consider persistent autonomous behaviors where a cyclic motion must be executed by the robot. The human operator is in charge of modifying online some geometric properties of the desired path. This is then autonomously processed by the robot in order to produce an actual path guaranteeing: i) tracking feasibility, ii) collision avoidance with obstacles, iii) closeness to the desired path set by the human operator, and iv) proximity to some points of interest. A force feedback is implemented to inform the human operator of the global deformation of the path rather than using the classical mismatch between desired and executed motion commands. Physically-based simulations, with human/hardware-in-the-loop and a quadrotor UAV as robotic platform, demonstrate the feasibility of the method.

This paper describes the mechanical and control design of the new 7-DOF CyberMotion Simulator, a redundant industrial manipulator arm consisting of a standard 6 DOF anthropomorphic manipulator plus an actuated cabin attached to the end-effector. Contrarily to Stewart platforms, an industrial manipulator offers several advantages when used as motion simulator: larger motion envelope, higher dexterity, and possibility to realize any end-effector posture within the workspace. In addition to this, the new actuated cabin acts as an additional joint and provides the needed kinematic redundancy to cope with the robot actuator and joint range constraints, which in general can significantly deteriorate the desired motion cues the robot is reproducing. In particular, we will show that, by suitably exploiting the redundancy better results can be obtained in reproducing sustained acceleration cues, a relevant problem when implementing vehicle simulators.

We present a decentralized system for the bilateral teleoperation of groups of UAVs which only relies on relative bearing measurements, i.e., without the need of distance information or global localization. The properties of a 3D bearing-formation are analyzed, and a minimal set of bearings needed for its definition is provided. We also design a novel decentralized formation control almost globally convergent and able to maintain bounded and non-vanishing inter-distances among the agents despite the absence of direct distance measurements. Furthermore, we develop a multi-master/ multi-slave teleoperation setup in order to control the overall behavior of the group and to convey to the human operator suitable force cues, while ensuring stability in presence of delays and packet losses over the master-slave communication channel. The theoretical framework is validated by means of extensive human/hardware in-the-loop simulations using two force-feedback devices and a group of quadrotors.

This paper considers the problem of realizing a 6-DOF closed-loop motion simulator by exploiting an anthropomorphic serial manipulator as motion platform. Contrary to standard Stewart platforms, an industrial anthropomorphic manipulator offers a considerably larger motion envelope and higher dexterity that let envisage it as a viable and superior alternative. Our work is divided in two papers. In this Part I, we discuss the main challenges in adopting a serial manipulator as motion platform, and thoroughly analyze one key issue: the design of a suitable inverse kinematics scheme for online motion reproduction. Experimental results are proposed to analyze the effectiveness of our approach. Part II [1] will address the design of a motion cueing algorithm tailored to the robot kinematics, and will provide an experimental evaluation on the chosen scenario: closed-loop simulation of a Formula 1 racing car.

This paper, divided in two Parts, considers the problem of realizing a 6-DOF closed-loop motion simulator by exploiting an anthropomorphic serial manipulator as motion platform. After having proposed a suitable inverse kinematics scheme in Part I [1], we address here the other key issue, i.e., devising a motion cueing algorithm tailored to the specific robot motion envelope. An extension of the well-known classical washout filter designed in cylindrical coordinates will provide an effective solution to this problem. The paper will then present a thorough experimental evaluation of the overall architecture (inverse kinematics + motion cueing) on the chosen scenario: closed-loop simulation of a Formula 1 racing car. This will prove the feasibility of our approach in fully exploiting the robot motion capabilities as a motion simulator.