The aim of this project is to create a low-cost device capable of acquiring and processing images, and of the development of an application that allows, through the collected images, a census and an analysis of the attributes of the persons detected in the images, taking advantage from the already existing online services for computer vision, such as those offered by Microsoft Azure Cognitive Service. In order to be able to profile customers, sample and visualize their data through statistical analysis, we use an accurate machine learning algorithm offered by this cloud service. The data are shown through a dedicated platform, generated with Microsoft PowerBi application, which allows we to easily recover the data collected in the Azure cloud databases, as well as easily manage the display of data by creating unique dashboards. The device selected to implement the solution has an ARM-type processor, which means Advanced RISC Machine. The main feature of the device is summarized in the latter acronym: Reduced Risc Instructions. This kind of processors is ideal for creating a simple and linear architecture. Currently there are numerous suppliers of this technology on the market, and among all, the RaspberryPI3 device has been selected for its performance and its decidedly low-cost price. II An operating system can be installed on this device, to manage it as if it were a full-fledged computer, taking into consideration that the application will have to interact with the Microsoft Azure services both for cognitive services and data storage. It is therefore convenient to choose the operating system that best fits to accomplish this purpose: Windows 10 IoT Core responds optimally to these needs. The only applications supported by this operating system are UWP (Universal Windows Platform) applications. They are special applications that can be installed on any Windows device: Hololens, Xbox, Mobile, Surface Hub, Mobile, PC and also on IoT devices, essential feature for the realization of the solution. The application interacts with REST-API calls, with the cognitive services offered by Azure, and the storage of metadata in the cloud following a re-elaboration of the same to optimize the face attributes collected.
Lo scopo di questo progetto è quello di realizzare un dispositivo a basso costo in grado di acquisire ed elaborare immagini, e dello sviluppo di un’applicazione che permetta mediante le immagini raccolte, un censimento e un'analisi degli attributi delle persone rilevate nelle immagini, sfruttando servizi online di computer vision già esistenti, come quelli offerti da Microsoft Azure Cognitive Service. Al fine di poter profilare la clientela, campionarla e visualizzarne i dati attraverso analisi statistiche, vengono sfruttati accurati algoritmi di machine learning offerti da questo servizio cloud. I dati vengono mostrati attraverso una piattaforma dedicata, generata con l'ausilio di Microsoft PowerBi, che permette facilmente di recuperare i dati collezionati nei database cloud di Azure, oltre che di gestire agevolmente la visualizzazione dei dati tramite la creazione di apposite dashboard. Il dispositivo selezionato per realizzare la soluzione ha un processore di tipo ARM, Advanced RISC Machine. La caratteristica principale del dispositivo si riassume in quest'ultimo acronimo: Reduced Risc Instructions. Questo genere di processori è l'ideale per realizzare un'architettura semplice e lineare. Attualmente sul mercato vi sono numerosi fornitori di questa tecnologia, e tra tutti, è stato selezionato il dispositivo RaspberryPI3 per le sue performance e il suo prezzo decisamente a basso costo. V Su questo dispositivo può essere installato un sistema operativo per poterlo gestire come se fosse un computer a tutti gli effetti, tenendo in considerazione che l'applicazione dovrà interagire con i servizi di Microsoft Azure sia per quanto riguarda i servizi cognitivi che la memorizzazione dei dati. È dunque conveniente scegliere il sistema operativo che si adatta al meglio per realizzare tale scopo: Windows 10 IoT Core risponde ottimamente a queste esigenze. Le uniche applicazioni supportate da questo sistema operativo sono le applicazioni di tipo UWP (Universal Windows Platform). Esse sono applicazioni particolari che possono essere installate su qualsiasi dispositivo Windows: Hololens, Xbox, Mobile, Surface Hub, Mobile, PC e anche sui dispositivi IoT, caratteristica indispensabile per la realizzazione della soluzione. L'applicazione interagisce attraverso chiamate REST-API, con i servizi cognitivi offerti da Azure ed il salvataggio dei metadati nel cloud a seguito di una rielaborazione degli stessi per ottimizzare al meglio le informazioni raccolte.
Design and implementation of a low-cost hardware and software platform for the profiling and analysis of store customers
FANTINI, ALESSANDRO
2017/2018
Abstract
The aim of this project is to create a low-cost device capable of acquiring and processing images, and of the development of an application that allows, through the collected images, a census and an analysis of the attributes of the persons detected in the images, taking advantage from the already existing online services for computer vision, such as those offered by Microsoft Azure Cognitive Service. In order to be able to profile customers, sample and visualize their data through statistical analysis, we use an accurate machine learning algorithm offered by this cloud service. The data are shown through a dedicated platform, generated with Microsoft PowerBi application, which allows we to easily recover the data collected in the Azure cloud databases, as well as easily manage the display of data by creating unique dashboards. The device selected to implement the solution has an ARM-type processor, which means Advanced RISC Machine. The main feature of the device is summarized in the latter acronym: Reduced Risc Instructions. This kind of processors is ideal for creating a simple and linear architecture. Currently there are numerous suppliers of this technology on the market, and among all, the RaspberryPI3 device has been selected for its performance and its decidedly low-cost price. II An operating system can be installed on this device, to manage it as if it were a full-fledged computer, taking into consideration that the application will have to interact with the Microsoft Azure services both for cognitive services and data storage. It is therefore convenient to choose the operating system that best fits to accomplish this purpose: Windows 10 IoT Core responds optimally to these needs. The only applications supported by this operating system are UWP (Universal Windows Platform) applications. They are special applications that can be installed on any Windows device: Hololens, Xbox, Mobile, Surface Hub, Mobile, PC and also on IoT devices, essential feature for the realization of the solution. The application interacts with REST-API calls, with the cognitive services offered by Azure, and the storage of metadata in the cloud following a re-elaboration of the same to optimize the face attributes collected.| File | Dimensione | Formato | |
|---|---|---|---|
|
fantini_thesis.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
2.52 MB
Formato
Adobe PDF
|
2.52 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/142107