Search

This paper presents a novel solution to automatically count vehicles in a parking lot using images captured by smart cameras. Unlike most of the literature on this task, which focuses on the analysis of single images, this paper proposes the use of multiple visual sources to monitor a wider parking area from different perspectives. The proposed multi-camera system is capable of automatically estimating the number of cars present in the entire parking lot directly on board the edge devices. It comprises an on-device deep learning-based detector that locates and counts the vehicles from the captured images and a decentralized geometric-based approach that can analyze the inter-camera shared areas and merge the data acquired by all the devices. We conducted the experimental evaluation on an extended version of the CNRPark-EXT dataset, a collection of images taken from the parking lot on the campus of the National Research Council (CNR) in Pisa, Italy. We show that our system is robust and takes advantage of the redundant information deriving from the different cameras, improving the overall performance without requiring any extra geometrical information of the monitored scene. • Smart mobility is crucial for smart cities and traffic-related issues. • We introduce a multi-camera system able to count cars from images of parking areas. • We combine a deep learning-based technique and a decentralized geometric approach. • All the algorithms run on the edge devices reducing the traffic on the network. • Our solution benefits from redundant information from different data sources. [ABSTRACT FROM AUTHOR]

Abstract: In recent years, the emerging diffusion of peer-to-peer networks is going beyond the single-domain paradigm like, for instance, the mono-thematic file sharing one (e.g. Napster for music). Peers are more and more heterogeneous data sources which need to share data with commercial, educational, and/or collaboration purposes, just to mention a few. Moreover, in current information processing applications data cannot be meaningfully searched by precise database queries that would return exact matches (e.g. when dealing with multimedia, proteomic, statistical data). In this paper we move a step towards multi-domain multi-type data sharing systems by introducing an advanced technological infrastructure which enables users to meet these new emerging needs. A fundamental issue in this context is data heterogeneity, which is pervasive and intrinsically present both at intensional level where, due to peers’ autonomy, different semantic descriptions of the available information are provided, and at extensional level, where multiple data types can coexist, also including content-based searchable data types such as multimedia data. Our proposal relies on a Peer Data Management Systems (PDMS) framework to present innovative network organization and query routing mechanisms which exploit both peers’ data description and data content to achieve effective and efficient network management and data retrieval in such a context. The validity of our proposal is demonstrated by an absolutely satisfactory experimental evaluation on a real setting. [ABSTRACT FROM AUTHOR]

While triggers have become a classical ingredient of relational database systems, research in active databases is aiming at extending the functionality and expressive power of active rules beyond the scope of relational triggers. One of the most important current trend concerns the support of detached active rules, i.e., of rules which are executed as separate transactions, running Outside of the scope of the transaction which generates the triggering event. Detached rules have important applications in workflow management and global integrity maintenance across transactions. One of the main issues in designing the rule engine for detached rules is determining their optimal scheduling. In this paper, we study the performance of a detached rule scheduler whose objective is to minimize the interference of detached rule execution with regard to the normal transactional Icad. This objective is achieved by executing detached rules at given periods of time aha by assigning them a fixed amount of dedicated threads; we study the performance of the scheduler relative to the two most critical design parameters, the frequency of execution of the scheduler, and the number of dedicated execution threads. [ABSTRACT FROM AUTHOR]

Introduces queueing network models for the performance analysis of Single Program Multiple Data applications. Basis of the models; Impact of different hardware and software surface components on the performance; Applicability of the models in anticipating the performance behavior of a parallel application.

Approximate Nearest Neighbor (ANN) search is a prevalent paradigm for searching intrinsically high dimensional objects in large-scale data sets. Recently, the permutation-based approach for ANN has attracted a lot of interest due to its versatility in being used in the more general class of metric spaces. In this approach, the entire database is ranked by a permutation distance to the query. Typically, permutations allow the efficient selection of a candidate set of results, but typically to achieve high recall or precision this set has to be reviewed using the original metric and data. This can lead to a sizeable percentage of the database being recalled, along with many expensive distance calculations. To reduce the number of metric computations and the number of database elements accessed, we propose here a re-ranking based on a local embedding using the nSimplex projection. The nSimplex projection produces Euclidean vectors from objects in metric spaces which possess the n-point property. The mapping is obtained from the distances to a set of reference objects, and the original metric can be lower bounded and upper bounded by the Euclidean distance of objects sharing the same set of references. Our approach is particularly advantageous for extensive databases or expensive metric function. We reuse the distances computed in the permutations in the first stage, and hence the memory footprint of the index is not increased. An extensive experimental evaluation of our approach is presented, demonstrating excellent results even on a set of hundreds of millions of objects. • Novel methods for refining results of permutation-based Nearest Neighbors search. • The proposed methods exploit pivot-based data embeddings. • The original data set is not accessed at query time as done by other refining methods. • Extensive experimental evaluation showing great improvements in the effectiveness. [ABSTRACT FROM AUTHOR]

In this paper, we investigate Hebbian learning strategies applied to Convolutional Neural Network (CNN) training. We consider two unsupervised learning approaches, Hebbian Winner-Takes-All (HWTA), and Hebbian Principal Component Analysis (HPCA). The Hebbian learning rules are used to train the layers of a CNN in order to extract features that are then used for classification, without requiring backpropagation (backprop). Experimental comparisons are made with state-of-the-art unsupervised (but backprop-based) Variational Auto-Encoder (VAE) training. For completeness,we consider two supervised Hebbian learning variants (Supervised Hebbian Classifiers—SHC, and Contrastive Hebbian Learning—CHL), for training the final classification layer, which are compared to Stochastic Gradient Descent training. We also investigate hybrid learning methodologies, where some network layers are trained following the Hebbian approach, and others are trained by backprop. We tested our approaches on MNIST, CIFAR10, and CIFAR100 datasets. Our results suggest that Hebbian learning is generally suitable for training early feature extraction layers, or to retrain higher network layers in fewer training epochs than backprop. Moreover, our experiments show that Hebbian learning outperforms VAE training, with HPCA performing generally better than HWTA. [ABSTRACT FROM AUTHOR]

Recent work on sample efficient training of Deep Neural Networks (DNNs) proposed a semi-supervised methodology based on biologically inspired Hebbian learning, combined with traditional backprop-based training. Promising results were achieved on various computer vision benchmarks, in scenarios of scarce labeled data availability. However, current Hebbian learning solutions can hardly address large-scale scenarios due to their demanding computational cost. In order to tackle this limitation, in this contribution, we investigate a novel solution, named FastHebb (FH), based on the reformulation of Hebbian learning rules in terms of matrix multiplications, which can be executed more efficiently on GPU. Starting from Soft-Winner-Takes-All (SWTA) and Hebbian Principal Component Analysis (HPCA) learning rules, we formulate their improved FH versions: SWTA-FH and HPCA-FH. We experimentally show that the proposed approach accelerates training speed up to 70 times, allowing us to gracefully scale Hebbian learning experiments on large datasets and network architectures such as ImageNet and VGG. [ABSTRACT FROM AUTHOR]

We propose to address the issue of sample efficiency, in Deep Convolutional Neural Networks (DCNN), with a semi-supervised training strategy that combines Hebbian learning with gradient descent: all internal layers (both convolutional and fully connected) are pre-trained using an unsupervised approach based on Hebbian learning, and the last fully connected layer (the classification layer) is trained using Stochastic Gradient Descent (SGD). In fact, as Hebbian learning is an unsupervised learning method, its potential lies in the possibility of training the internal layers of a DCNN without labels. Only the final fully connected layer has to be trained with labeled examples. We performed experiments on various object recognition datasets, in different regimes of sample efficiency, comparing our semi-supervised (Hebbian for internal layers + SGD for the final fully connected layer) approach with end-to-end supervised backprop training, and with semi-supervised learning based on Variational Auto-Encoder (VAE). The results show that, in regimes where the number of available labeled samples is low, our semi-supervised approach outperforms the other approaches in almost all the cases. • Semi-supervised approach combining Hebbian learning with SGD. • Unsupervised pre-training phase based on Hebbian PCA. • Followed by a supervised training phase based on SGD. • Comparison with end-to-end supervised backprop and unsupervised VAE pre-training. • Experiments show that our approach (Hebbian + SGD) outperforms the others. [ABSTRACT FROM AUTHOR]

Permutation-based Indexing (PBI) approaches have been proven to be particularly effective for conducting large-scale approximate metric searching. These methods rely on the idea of transforming the original metric objects into permutation representations, which can be efficiently indexed using data structures such as inverted files. The standard conceptualization of permutation associated with a metric object involves only the use of object distances and their relative orders from a set of anchors called pivots. In this paper, we generalized this definition in order to enlarge the class of permutation representations that can be used by PBI approaches. In particular, we introduced the concept of permutation induced by a space transformation and a sorting function, and we investigated which properties these transformations should possess to produce permutations that are effective for metric search. Furthermore, as a practical outcome, we defined a new type of permutation representation that is calculated using distances from pairs of pivots. This proposed technique allowed us to produce longer permutations than traditional ones for the same number of object-pivot distance calculations. The advantage lies in the fact that when longer permutations are employed, the use of inverted files built on permutation prefixes leads to greater efficiency in the search phase. • Permutation-based Indexing methods are effective for large-scale metric search. • Definition of permutations induced by a space transformation. • Study of space transformation properties to induce good permutations for metric search. • Novel permutations calculated using both object-pivot and pivot-pivot distances. [ABSTRACT FROM AUTHOR]

In non-terrestrial networks, where low Earth orbit satellites and user equipment move relative to each other, line-of-sight tracking and adapting to channel state variations due to endpoint movements are a major challenge. Therefore, continuous line-of-sight estimation and channel impairment compensation are crucial for user equipment to access a satellite and maintain connectivity. In this paper, we propose a framework based on actor-critic reinforcement learning for traffic scheduling in non-terrestrial networks scenario where the channel state is non-stationary due to the variability of the line of sight, which depends on the current satellite elevation. We deploy the framework as an agent in a multipath routing scheme where the user equipment can access more than one satellite simultaneously to improve link reliability and throughput. We investigate how the agent schedules traffic in multiple satellite links by adopting policies that are evaluated by an actor-critic reinforcement learning approach. The agent continuously trains its model based on variations in satellite elevation angles, handovers, and relative line-of-sight probabilities. We compare the agent's retraining time with the satellite visibility intervals to investigate the effectiveness of the agent's learning rate. We carry out performance analysis while considering the dense urban area of Paris, where high-rise buildings significantly affect the line of sight. The simulation results show how the learning agent selects the scheduling policy when it is connected to a pair of satellites. The results also show that the retraining time of the learning agent is up to 0.1 t i m e s the satellite visibility time at given elevations, which guarantees efficient use of satellite visibility. [ABSTRACT FROM AUTHOR]

The automatic detection of violent actions in public places through video analysis is difficult because the employed Artificial Intelligence-based techniques often suffer from generalization problems. Indeed, these algorithms hinge on large quantities of annotated data and usually experience a drastic drop in performance when used in scenarios never seen during the supervised learning phase. In this paper, we introduce and publicly release the Bus Violence benchmark, the first large-scale collection of video clips for violence detection on public transport, where some actors simulated violent actions inside a moving bus in changing conditions, such as the background or light. Moreover, we conduct a performance analysis of several state-of-the-art video violence detectors pre-trained with general violence detection databases on this newly established use case. The achieved moderate performances reveal the difficulties in generalizing from these popular methods, indicating the need to have this new collection of labeled data, beneficial for specializing them in this new scenario. [ABSTRACT FROM AUTHOR]

• Same-different problems can be solved using modern convolutional networks. • Residual connections seems not to have a primary role in the convergence. • Older networks (AlexNet and VGG) are strained, and cannot converge in many cases. • We reach SOTA results and super-human performances on three out of four SVRT tasks. • We demonstrate zero-shot generalization abilities to similar problems. Deep learning demonstrated major abilities in solving many kinds of different real-world problems in computer vision literature. However, they are still strained by simple reasoning tasks that humans consider easy to solve. In this work, we probe current state-of-the-art convolutional neural networks on a difficult set of tasks known as the same-different problems. All the problems require the same prerequisite to be solved correctly: understanding if two random shapes inside the same image are the same or not. With the experiments carried out in this work, we demonstrate that residual connections, and more generally the skip connections, seem to have only a marginal impact on the learning of the proposed problems. In particular, we experiment with DenseNets, and we examine the contribution of residual and recurrent connections in already tested architectures, ResNet-18, and CorNet-S respectively. Our experiments show that older feed-forward networks, AlexNet and VGG, are almost unable to learn the proposed problems, except in some specific scenarios. We show that recently introduced architectures can converge even in the cases where the important parts of their architecture are removed. We finally carry out some zero-shot generalization tests, and we discover that in these scenarios residual and recurrent connections can have a stronger impact on the overall test accuracy. On four difficult problems from the SVRT dataset, we can reach state-of-the-art results with respect to the previous approaches, obtaining super-human performances on three of the four problems. [ABSTRACT FROM AUTHOR]

A smart camera is a vision system capable of extracting application-specific information from the captured images. The paper proposes a decentralized and efficient solution for visual parking lot occupancy detection based on a deep Convolutional Neural Network (CNN) specifically designed for smart cameras. This solution is compared with state-of-the-art approaches using two visual datasets: PKLot, already existing in literature, and CNRPark-EXT. The former is an existing dataset, that allowed us to exhaustively compare with previous works. The latter dataset has been created in the context of this research, accumulating data across various seasons of the year, to test our approach in particularly challenging situations, exhibiting occlusions, and diverse and difficult viewpoints. This dataset is public available to the scientific community and is another contribution of our research. Our experiments show that our solution outperforms and generalizes the best performing approaches on both datasets. The performance of our proposed CNN architecture on the parking lot occupancy detection task, is comparable to the well-known AlexNet, which is three orders of magnitude larger. [ABSTRACT FROM AUTHOR]

In many working and recreational activities, there are scenarios where both individual and collective safety have to be constantly checked and properly signaled, as occurring in dangerous workplaces or during pandemic events like the recent COVID-19 disease. From wearing personal protective equipment to filling physical spaces with an adequate number of people, it is clear that a possibly automatic solution would help to check compliance with the established rules. Based on an off-the-shelf compact and low-cost hardware, we present a deployed real use-case embedded system capable of perceiving people's behavior and aggregations and supervising the appliance of a set of rules relying on a configurable plug-in framework. Working on indoor and outdoor environments, we show that our implementation of counting people aggregations, measuring their reciprocal physical distances, and checking the proper usage of protective equipment is an effective yet open framework for monitoring human activities in critical conditions. • Monitoring compliance with safety rules is crucial in critical environments. • Presenting a Computer Vision-based AI-assisted system to monitor human activities. • Modular architecture for pedestrian detection, counting, distancing, and PPE detection. • Two novel datasets of PPE detection and overall system evaluation in a real scenario. • State-of-the-art trained models in a deployed real use-case scenario in Pisa, Italy. [ABSTRACT FROM AUTHOR]

• Counting cells in microscopy images is a crucial step to diagnose many diseases. • Non-trivial patterns induce weak labels in annotation procedures, even with experts. • A two-stage cell counting method for weakly-labeled data settings is proposed. • We localize and score objects to correlate prediction scores and raters' agreements. Our method achieves lower counting errors across multiple ground-truth settings. [Display omitted] Exploiting well-labeled training sets has led deep learning models to astonishing results for counting biological structures in microscopy images. However, dealing with weak multi-rater annotations, i.e., when multiple human raters disagree due to non-trivial patterns, remains a relatively unexplored problem. More reliable labels can be obtained by aggregating and averaging the decisions given by several raters to the same data. Still, the scale of the counting task and the limited budget for labeling prohibit this. As a result, making the most with small quantities of multi-rater data is crucial. To this end, we propose a two-stage counting strategy in a weakly labeled data scenario. First, we detect and count the biological structures; then, in the second step, we refine the predictions, increasing the correlation between the scores assigned to the samples and the raters' agreement on the annotations. We assess our methodology on a novel dataset comprising fluorescence microscopy images of mice brains containing extracellular matrix aggregates named perineuronal nets. We demonstrate that we significantly enhance counting performance, improving confidence calibration by taking advantage of the redundant information characterizing the small sets of available multi-rater data. [ABSTRACT FROM AUTHOR]

Robotic ecologies are systems made out of several robotic devices, including mobile robots, wireless sensors and effectors embedded in everyday environments, where they cooperate to achieve complex tasks. This paper demonstrates how endowing robotic ecologies with information processing algorithms such as perception, learning, planning, and novelty detection can make these systems able to deliver modular, flexible, manageable and dependable Ambient Assisted Living (AAL) solutions. Specifically, we show how the integrated and self-organising cognitive solutions implemented within the EU project RUBICON (Robotic UBIquitous Cognitive Network) can reduce the need of costly pre-programming and maintenance of robotic ecologies. We illustrate how these solutions can be harnessed to (i) deliver a range of assistive services by coordinating the sensing & acting capabilities of heterogeneous devices, (ii) adapt and tune the overall behaviour of the ecology to the preferences and behaviour of its inhabitants, and also (iii) deal with novel events, due to the occurrence of new user׳s activities and changing user׳s habits. [ABSTRACT FROM AUTHOR]

Pedestrian detection through Computer Vision is a building block for a multitude of applications. Recently, there has been an increasing interest in convolutional neural network-based architectures to execute such a task. One of these supervised networks' critical goals is to generalize the knowledge learned during the training phase to new scenarios with different characteristics. A suitably labeled dataset is essential to achieve this purpose. The main problem is that manually annotating a dataset usually requires a lot of human effort, and it is costly. To this end, we introduce ViPeD (Virtual Pedestrian Dataset), a new synthetically generated set of images collected with the highly photo-realistic graphical engine of the video game GTA V (Grand Theft Auto V), where annotations are automatically acquired. However, when training solely on the synthetic dataset, the model experiences a Synthetic2Real domain shift leading to a performance drop when applied to real-world images. To mitigate this gap, we propose two different domain adaptation techniques suitable for the pedestrian detection task, but possibly applicable to general object detection. Experiments show that the network trained with ViPeD can generalize over unseen real-world scenarios better than the detector trained over real-world data, exploiting the variety of our synthetic dataset. Furthermore, we demonstrate that with our domain adaptation techniques, we can reduce the Synthetic2Real domain shift, making the two domains closer and obtaining a performance improvement when testing the network over the real-world images. [ABSTRACT FROM AUTHOR]

Smart Home has gained widespread attention due to its flexible integration into everyday life. Pervasive sensing technologies are used to recognize and track the activities that people perform during the day, and to allow communication and cooperation of physical objects. Usually, the available infrastructures and applications leveraging these smart environments have a critical impact on the overall cost of the Smart Home construction, require to be preferably installed during the home construction and are still not user-centric. In this paper, we propose a low cost, easy to install, user-friendly, dynamic and flexible infrastructure able to perform runtime resources management by decoupling the different levels of control rules. The basic idea relies on the usage of off-the-shelf sensors and technologies to guarantee the regular exchange of critical information, without the necessity from the user to develop accurate models for managing resources or regulating their access/usage. This allows us to simplify the continuous updating and improvement, to reduce the maintenance effort and to improve residents’ living and security. A first validation of the proposed infrastructure on a case study is also presented. [ABSTRACT FROM AUTHOR]