Search

We present X-ray and multi-band optical observations of the afterglow and host galaxy of GRB 180418A, discovered by ${\it Swift}$/BAT and ${\it Fermi}$/GBM. We present a reanalysis of the GBM and BAT data deriving durations of the prompt emission of $T_{90}\approx$2.56s and $\approx$1.90s, respectively. Modeling the ${\it Fermi}$/GBM catalog of 1405 bursts (2008-2014) in the Hardness-$T_{90}$ plane, we obtain a probability of $\approx$60% that GRB 180418A is a short-hard burst. From a combination of ${\it Swift}$/XRT and ${\it Chandra}$ observations, the X-ray afterglow is detected to $\approx$38.5 days after the burst, and exhibits a single power-law decline with $F_{\rm X} \propto t^{-0.98}$. Late-time Gemini observations reveal a faint r$\approx$25.69 mag host galaxy at an angular offset of $\approx$0.16''. At the likely redshift range of z$\approx$1-2.25, we find that the X-ray afterglow luminosity of GRB 180418A is intermediate between short and long GRBs at all epochs during which there is contemporaneous data, and that GRB 180418A lies closer to the $E_{\gamma,{\rm peak}}-E_{\gamma,{\rm iso}}$ correlation for short GRBs. Modeling the multi-wavelength afterglow with the standard synchrotron model, we derive the burst explosion properties and find a jet opening angle of $\theta_{\rm j} \gtrsim 9-14^{\circ}$. If GRB 180418A is a short GRB that originated from a neutron star merger, it has one of the brightest and longest-lived afterglows along with an extremely faint host galaxy. If instead the event is a long GRB that originated from a massive star collapse, it has among the lowest luminosity afterglows, and lies in a peculiar space in terms of the Hardness-$T_{90}$ and $E_{\gamma,{\rm peak}}-E_{\gamma,{\rm iso}}$ planes., Comment: 26 pages, 9 figures, 3 tables (Accepted, ApJ)

We present multiwavelength modeling of the afterglow from the long gamma-ray burst GRB 160625B using Markov Chain Monte Carlo (MCMC) techniques of the afterglowpy Python package. GRB 160625B is an extremely bright burst with a rich set of observations spanning from radio to gamma-ray frequencies. These observations range from ~0.1 days to >1000 days, thus making this event extremely well-suited to such modeling. In this work we compare top-hat and Gaussian jet structure types in order to find best fit values for the GRB jet collimation angle, viewing angle, and other physical parameters. We find that a Gaussian-shaped jet is preferred (2.7-5.3 sigma) over the traditional top-hat model. Our estimate for the opening angle of the burst ranges from 1.26 to 3.90 degrees, depending on jet shape model. We also discuss the implications that assumptions on jet shape, viewing angle, and particularly the participation fraction of electrons have on the final estimation of GRB intrinsic energy release and the resulting energy budget of the relativistic outflow. Most notably, allowing the participation fraction to vary results in an estimated total relativistic energy of ~$10^{53}$ erg. This is two orders of magnitude higher than when the total fraction is assumed to be unity, thus this parameter has strong relevance for placing constraints on long GRB central engines, details of the circumburst media, and host environment., Comment: 29 pages, 10 figures, accepted by ApJ

We present the discovery of the radio afterglow and near-infrared (NIR) counterpart of the Swift short GRB 200522A, located at a small projected offset of $\approx 1$ kpc from the center of a young, star-forming host galaxy at $z=0.5536$. The radio and X-ray luminosities of the afterglow are consistent with those of on-axis cosmological short GRBs. The NIR counterpart, revealed by our HST observations at a rest-frame time of $\approx2.3$ days, has a luminosity of $\approx (1.3-1.7) \times 10^{42}$ erg s$^{-1}$. This is substantially lower than on-axis short GRB afterglow detections, but is a factor of $\approx 8$-$17$ more luminous than the kilonova of GW170817, and significantly more luminous than any kilonova candidate for which comparable observations exist. The combination of the counterpart's color ($i-y = -0.08\pm 0.21$; rest-frame) and luminosity cannot be explained by standard radioactive heating alone. We present two scenarios to interpret the broad-band behavior of GRB 200522A: a synchrotron forward shock with a luminous kilonova (potentially boosted by magnetar energy deposition), or forward and reverse shocks from a $\approx14^{\circ}$, relativistic ($\Gamma_0 \gtrsim 80$) jet. Models which include a combination of enhanced radioactive heating rates, low-lanthanide mass fractions, or additional sources of heating from late-time central engine activity may provide viable alternate explanations. If a stable magnetar was indeed produced in GRB 200522A, we predict that late-time radio emission will be detectable starting $\approx 0.3$-$6$ years after the burst for a deposited energy of $\approx 10^{53}$ erg. Counterparts of similar luminosity to GRB 200522A associated with gravitational wave events will be detectable with current optical searches to $\approx\!250$ Mpc., Comment: 33 pages, 13 figures, 5 tables. Submitted to AAS Journals

In this document, we report our proposal for modeling the risk of possible contagiousity in a given area monitored by RGB cameras where people freely move and interact. Our system, called Inter-Homines, evaluates in real-time the contagion risk in a monitored area by analyzing video streams: it is able to locate people in 3D space, calculate interpersonal distances and predict risk levels by building dynamic maps of the monitored area. Inter-Homines works both indoor and outdoor, in public and private crowded areas. The software is applicable to already installed cameras or low-cost cameras on industrial PCs, equipped with an additional embedded edge-AI system for temporary measurements. From the AI-side, we exploit a robust pipeline for real-time people detection and localization in the ground plane by homographic transformation based on state-of-the-art computer vision algorithms; it is a combination of a people detector and a pose estimator. From the risk modeling side, we propose a parametric model for a spatio-temporal dynamic risk estimation, that, validated by epidemiologists, could be useful for safety monitoring the acceptance of social distancing prevention measures by predicting the risk level of the scene.

Embodied AI has been recently gaining attention as it aims to foster the development of autonomous and intelligent agents. In this paper, we devise a novel embodied setting in which an agent needs to explore a previously unknown environment while recounting what it sees during the path. In this context, the agent needs to navigate the environment driven by an exploration goal, select proper moments for description, and output natural language descriptions of relevant objects and scenes. Our model integrates a novel self-supervised exploration module with penalty, and a fully-attentive captioning model for explanation. Also, we investigate different policies for selecting proper moments for explanation, driven by information coming from both the environment and the navigation. Experiments are conducted on photorealistic environments from the Matterport3D dataset and investigate the navigation and explanation capabilities of the agent as well as the role of their interactions., Comment: ICPR 2020

We present the discovery of the optical afterglow and host galaxy of the {\it Swift} short-duration gamma-ray burst, GRB\,181123B. Observations with Gemini-North starting at $\approx 9.1$~hr after the burst reveal a faint optical afterglow with $i\approx25.1$~mag, at an angular offset of 0.59 $\pm$ 0.16$''$ from its host galaxy. Using $grizYJHK$ observations, we measure a photometric redshift of the host galaxy of $z = 1.77^{+0.30}_{-0.17}$. From a combination of Gemini and Keck spectroscopy of the host galaxy spanning 4500-18000~\AA , we detect a single emission line at 13390~\AA, inferred as H$\beta$ at $z = 1.754 \pm 0.001$ and corroborating the photometric redshift. The host galaxy properties of GRB\,181123B are typical to those of other SGRB hosts, with an inferred stellar mass of $\approx 1.7 \times 10^{10}\,M_{\odot}$, mass-weighted age of $\approx 0.9$~Gyr and optical luminosity of $\approx 0.9L^{*}$. At $z=1.754$, GRB\,181123B is the most distant secure SGRB with an optical afterglow detection, and one of only three at $z>1.5$. Motivated by a growing number of high-$z$ SGRBs, we explore the effects of a missing $z>1.5$ SGRB population among the current {\it Swift} sample on delay time distribution models. We find that log-normal models with mean delay times of $\approx 4-6$~Gyr are consistent with the observed distribution, but can be ruled out to $95\%$ confidence with an additional $\approx1-5$~{\it Swift} SGRBs recovered at $z>1.5$. In contrast, power-law models with $\propto$ $t^{-1}$ are consistent with the redshift distribution and can accommodate up to $\approx30$ SGRBs at these redshifts. Under this model, we predict that $\approx 1/3$ of the current {\it Swift} population of SGRBs is at $z>1$. The future discovery or recovery of existing high-$z$ SGRBs will provide significant discriminating power on their delay time distributions, and thus their formation channels.

In this work we propose a deep learning pipeline to predict the visual future appearance of an urban scene. Despite recent advances, generating the entire scene in an end-to-end fashion is still far from being achieved. Instead, here we follow a two stages approach, where interpretable information is included in the loop and each actor is modelled independently. We leverage a per-object novel view synthesis paradigm; i.e. generating a synthetic representation of an object undergoing a geometrical roto-translation in the 3D space. Our model can be easily conditioned with constraints (e.g. input trajectories) provided by state-of-the-art tracking methods or by the user itself. This allows us to generate a set of diverse realistic futures starting from the same input in a multi-modal fashion. We visually and quantitatively show the superiority of this approach over traditional end-to-end scene-generation methods on CityFlow, a challenging real world dataset., Comment: Accepted at ICPR2020

Understanding human motion behaviour is a critical task for several possible applications like self-driving cars or social robots, and in general for all those settings where an autonomous agent has to navigate inside a human-centric environment. This is non-trivial because human motion is inherently multi-modal: given a history of human motion paths, there are many plausible ways by which people could move in the future. Additionally, people activities are often driven by goals, e.g. reaching particular locations or interacting with the environment. We address the aforementioned aspects by proposing a new recurrent generative model that considers both single agents' future goals and interactions between different agents. The model exploits a double attention-based graph neural network to collect information about the mutual influences among different agents and to integrate it with data about agents' possible future objectives. Our proposal is general enough to be applied to different scenarios: the model achieves state-of-the-art results in both urban environments and also in sports applications., Comment: Accepted at ICPR 2020

Anticipating human motion in crowded scenarios is essential for developing intelligent transportation systems, social-aware robots and advanced video surveillance applications. A key component of this task is represented by the inherently multi-modal nature of human paths which makes socially acceptable multiple futures when human interactions are involved. To this end, we propose a generative architecture for multi-future trajectory predictions based on Conditional Variational Recurrent Neural Networks (C-VRNNs). Conditioning mainly relies on prior belief maps, representing most likely moving directions and forcing the model to consider past observed dynamics in generating future positions. Human interactions are modeled with a graph-based attention mechanism enabling an online attentive hidden state refinement of the recurrent estimation. To corroborate our model, we perform extensive experiments on publicly-available datasets (e.g., ETH/UCY, Stanford Drone Dataset, STATS SportVU NBA, Intersection Drone Dataset and TrajNet++) and demonstrate its effectiveness in crowded scenes compared to several state-of-the-art methods., Comment: Accepted at Computer Vision and Image Understanding (CVIU)

The joint understanding of vision and language has been recently gaining a lot of attention in both the Computer Vision and Natural Language Processing communities, with the emergence of tasks such as image captioning, image-text matching, and visual question answering. As both images and text can be encoded as sets or sequences of elements -- like regions and words -- proper reduction functions are needed to transform a set of encoded elements into a single response, like a classification or similarity score. In this paper, we propose a novel fully-attentive reduction method for vision and language. Specifically, our approach computes a set of scores for each element of each modality employing a novel variant of cross-attention, and performs a learnable and cross-modal reduction, which can be used for both classification and ranking. We test our approach on image-text matching and visual question answering, building fair comparisons with other reduction choices, on both COCO and VQA 2.0 datasets. Experimentally, we demonstrate that our approach leads to a performance increase on both tasks. Further, we conduct ablation studies to validate the role of each component of the approach., Comment: ICPR 2020

In this paper we present a novel approach for bottom-up multi-person 3D human pose estimation from monocular RGB images. We propose to use high resolution volumetric heatmaps to model joint locations, devising a simple and effective compression method to drastically reduce the size of this representation. At the core of the proposed method lies our Volumetric Heatmap Autoencoder, a fully-convolutional network tasked with the compression of ground-truth heatmaps into a dense intermediate representation. A second model, the Code Predictor, is then trained to predict these codes, which can be decompressed at test time to re-obtain the original representation. Our experimental evaluation shows that our method performs favorably when compared to state of the art on both multi-person and single-person 3D human pose estimation datasets and, thanks to our novel compression strategy, can process full-HD images at the constant runtime of 8 fps regardless of the number of subjects in the scene. Code and models available at https://github.com/fabbrimatteo/LoCO ., Comment: CVPR 2020

Convolutional Neural Networks experience catastrophic forgetting when optimized on a sequence of learning problems: as they meet the objective of the current training examples, their performance on previous tasks drops drastically. In this work, we introduce a novel framework to tackle this problem with conditional computation. We equip each convolutional layer with task-specific gating modules, selecting which filters to apply on the given input. This way, we achieve two appealing properties. Firstly, the execution patterns of the gates allow to identify and protect important filters, ensuring no loss in the performance of the model for previously learned tasks. Secondly, by using a sparsity objective, we can promote the selection of a limited set of kernels, allowing to retain sufficient model capacity to digest new tasks.Existing solutions require, at test time, awareness of the task to which each example belongs to. This knowledge, however, may not be available in many practical scenarios. Therefore, we additionally introduce a task classifier that predicts the task label of each example, to deal with settings in which a task oracle is not available. We validate our proposal on four continual learning datasets. Results show that our model consistently outperforms existing methods both in the presence and the absence of a task oracle. Notably, on Split SVHN and Imagenet-50 datasets, our model yields up to 23.98% and 17.42% improvement in accuracy w.r.t. competing methods., Comment: CVPR 2020 (oral)