Guido Matrella

An Assistive Home Automation and Monitoring System

A versatile, reliable and inexpensive home automation system is presented, suited for assisting and monitoring elderly and disabled people in home daily living. A smart interface module has been fabricated, which allows for straightforward integration of a wide variety of devices in a standard LAN. A pilot installation provides successfully HW and SW validation.

Food Image Recognition Using Very Deep Convolutional Networks

We evaluated the effectiveness in classifying food images of a deep-learning approach based on the specifications of Googles image recognition architecture Inception. The architecture is a deep convolutional neural network (DCNN) having a depth of 54 layers. In this study, we fine-tuned this architecture for classifying food images from three well-known food image datasets: ETH Food-101, UEC FOOD 100, and UEC FOOD 256. On these datasets we achieved, respectively, 88.28%, 81.45%, and 76.17% as top-1 accuracy and 96.88%, 97.27%, and 92.58% as top-5 accuracy. To the best of our knowledge, these results significantly improve the best published results obtained on the same datasets, while requiring less computation power, since the number of parameters and the computational complexity are much smaller than the competitors?. Because of this, even if it is still rather large, the deep network based on this architecture appears to be at least closer to the requirements for mobile systems.

A Wireless Sensor Platform for Assistive Technology Applications

In this paper, the development of a prototypal wireless sensor platform is described, aimed at assisting elderly people and people with disabilities in their daily living activities at home. The wireless sensor network is embedded into a more general home control and monitoring network, from which it can borrow remote communication and supervision facilities, enhancing versatility and reliability. A wearable sensor has been developed, capable of smart recognition of abnormal gait and falls. Effective algorithms have been devised, to make the device suitable for low-power hardware implementation. After initial prototyping phases (based on microcontrollers and FPGA) VLSI synthesis has been carried out, to estimate actual silicon area and power consumption. Then, extensions of the approach have been foreseen, accounting for multiple sensor management. An example of a low cost embedded heartbeat monitor is discussed.

Hardware-oriented Adaptation of a Particle Swarm Optimization Algorithm for Object Detection

In this paper we propose a simplified, hardware-oriented algorithm for object detection, based on particle swarm optimization. Starting from an algorithm coded in a high-level language which has shown to perform well, both in terms of accuracy and of computation efficiency, the simplified version can be implemented on an FPGA. After describing the original algorithm, we describe how it has been simplified for hardware implementation. We show how the intrinsic modularity of the algorithm permits to define a general core, independent of the specific application, which implements object search, along with a simple application specific-module, which implements a problem-dependent fitness function. This makes the system easily reconfigurable when switching between different object detection applications. Finally, we show some examples of application of our algorithm and discuss about possible future developments.

Automatic diet monitoring: a review of computer vision and wearable sensor-based methods

Abstract Food intake and eating habits have a significant impact on people’s health. Widespread diseases, such as diabetes and obesity, are directly related to eating habits. Therefore, monitoring diet can be a substantial base for developing methods and services to promote healthy lifestyle and improve personal and national health economy. Studies have demonstrated that manual reporting of food intake is inaccurate and often impractical. Thus, several methods have been proposed to automate the process. This article reviews the most relevant and recent researches on automatic diet monitoring, discussing their strengths and weaknesses. In particular, the article reviews two approaches to this problem, accounting for most of the work in the area. The first approach is based on image analysis and aims at extracting information about food content automatically from food images. The second one relies on wearable sensors and has the detection of eating behaviours as its main goal.

Advanced active pixel architectures in standard CMOS technology

This paper aims at exploring and validating the adoption of standard fabrication processes for the realization of CMOS active pixel sensors, for particle detection purposes. The goal is to implement a single-chip, complete radiation sensor system, including on a CMOS integrated circuit the sensitive devices, read-out and signal processing circuits. A prototype chip (RAPS01) based on these principles has been already fabricated, and a chip characterization has been carried out; in particular, the evaluation of the sensitivity of the sensor response on the actual operating conditions was estimated, as well as the response uniformity. Optimization and tailoring of the sensor structures for High Energy Physics applications are being evaluated in the design of the next generation chip (RAPS02). Basic features of the new chip includes digitally configurable readout and multi-mode access (i.e., either sparse of line-scan readout).

The HELICOPTER Project: A Heterogeneous Sensor Network Suitable for Behavioral Monitoring

In this paper, the infrastructure supporting the HELICOPTER AAL-JP project is described. The project aims at introducing behavioral analysis features for early detection of age-related diseases: to this purpose, a heterogeneous sensor network has been designed and implemented, encompassing in the same vision environmental, wearable and clinical sensors. In order to make environmental sensors suitable for behavioral inference, the issue of activity tagging i.e., attribution to a given user of the action detected by the sensors needs to be tackled. Within the HELICOPTER scenario, cooperation between environmental and wearable sensors is exploited to this aim. Preliminary results offer encouraging perspectives: piloting phase, which will validate the approach on a larger scale, is close to start.

Advances in radiation active pixel sensors (RAPS) architectures

This work aims at exploring and validating the adoption of standard fabrication processes for the realization of CMOS active pixel sensor, for particle detection purposes. The goal is to implement a single-chip, complete radiation sensor system, including on a CMOS IC the sensitive devices, read-out and signal processing circuits. The possibility of including versatile and performing circuitry allows for the evaluation of innovative active pixel architectures, different read-out strategies, and complex data management algorithms. A prototype chip (RAPS01) based on these principles has been already fabricated, and a complete chip characterization has been carried out; in particular, the evaluation of the sensitivity of the sensor response on the actual operating conditions was estimated, as well as uniformity response analysis. Optimization and tailoring of the sensor structures for specific applications are being evaluated in the design of the next generation chip (RAPS02). In particular, sparse read-out approach and power consumption are considered, introducing some circuit improvement, and discussing the organization and design of a new architecture. Basic features of the new chip includes: digitally configurable readout, power-switching techniques, fault-tolerant circuitry, multi-mode access (i.e., either sparse of line-scan readout). Thanks to the intrinsic flexibility of CMOS design, perspective application different from HEP experiments, can be evaluated as well.

A New Approach to Image-Based Estimation of Food Volume

A balanced diet is the key to a healthy lifestyle and is crucial for preventing or dealing with many chronic diseases such as diabetes and obesity. Therefore, monitoring diet can be an effective way of improving people’s health. However, manual reporting of food intake has been shown to be inaccurate and often impractical. This paper presents a new approach to food intake quantity estimation using image-based modeling. The modeling method consists of three steps: firstly, a short video of the food is taken by the user’s smartphone. From such a video, six frames are selected based on the pictures’ viewpoints as determined by the smartphone’s orientation sensors. Secondly, the user marks one of the frames to seed an interactive segmentation algorithm. Segmentation is based on a Gaussian Mixture Model alongside the graph-cut algorithm. Finally, a customized image-based modeling algorithm generates a point-cloud to model the food. At the same time, a stochastic object-detection method locates a checkerboard used as size/ground reference. The modeling algorithm is optimized such that the use of six input images still results in an acceptable computation cost. In our evaluation procedure, we achieved an average accuracy of 92 % on a test set that includes images of different kinds of pasta and bread, with an average processing time of about 23 s.

An Ontology Designed for Supporting an AAL Experience in the Framework of the FOOD Project

This work focuses on the design and the implementation of an ontology devised to extract knowledge from an Ambient Assisted Living environment, equipped with an Home Automation System. In particular, the design of the ontology is been carried out taking as reference the experience of the FOOD project [1], funded in the framework of the European AAL Joint Programme [2]. In the pilot installations of the FOOD project, a set of sensors was installed in the kitchen of the users in order to monitor the behavior of the persons, aiming to recording their feeding habits. The main part of the performed activity is the design of the ontology itself and, in particular: the identification of most important conceptual classes, the way to describe the concept of time within the ontology, and the development of a relational database that enables the interfacing between the ontology and the actual data provided by the sensors and collected by the Home Automation System.