HP Solutions Creation & Development Services
HP SCDS
C/ Julia Morros 1, Edificio Zarzuela - Parque Tecnológico de León. 24005 - León (SPAIN)
ChronoStreetTourist is a mobile app that uses augmented reality to display historical images on a mobile device's camera when pointed at a building or location for which old images are available.
It was developed in previous Final Year Projects (FYP) during the 2021-22 and 2022-23 academic years (the latter by the co-director Ánder). Now, the goal is to further develop the ChronoStreetTourist project.
Once operational, ChronoStreetTourist project aims to evolve in different aspects:
2. Simplify architecture: The previous project had two components: a mobile app that displays augmented reality when viewing a monument and a web interface for managing images, users, etc.
a. Since the hosting is down, the goal is to eliminate the web interface and make the app function autonomously.
b. Implement an admin user role with access to all functionalities and a user role with limited access.
3. Geolocation: Integrate a map marking all the monuments included in the database. When clicking on these points on the map, the old photo would be displayed along with a link to a website with more information about it.
4. Push notifications: Notify the user if they approach within approximately 100 meters of any of the map points.
5. Additional improvements: If time permits, consider enhancements to the app’s aesthetics, QR code reading, etc. The student has the freedom to propose other improvements.
During the last two years, the Technological Observatory has developed the “Diaberse” project, an immersive virtual reality experience for diabetic education. Through videos and games, children and adults, especially those who are new to this pathology, are taught how to face different real life situations when you are diabetic.
In this extension, our diabetes consultation will be more interactive than ever, where our avatar, in addition to understanding voice commands, will be able to help us control our diabetes in an easy, intuitive and safe way. First of all, we will configure our avatar to be able to understand voice commands, which, once processed, and through an LLM that we can re-train or not, really understand what we want to know about diabetes and, through voice, will give us a result that can be complemented with more information, either text, images or videos. Thus, we will be able to have a conversation with our avatar whose purpose will be the control of diabetes, either type 1 or type 2.
For the realization of this project, the student will be provided with access to all the code of previous editions.
One of the problems today is that for many people, the prospectus is a complicated document to read, to find relevant information, or it gets lost and then some information is needed. The Spanish Agency of Medicines and Health Products (AEMPS) has made the CIMA (Center for Information on Medicines) publicly available.
The objective of this project is to use the information available in CIMA to offer the user a simple experience of managing the information of a medication. For this, the mobile application must be able to recognize a medication through the camera (photo of the box), then display the prospectus and be able to summarize it with useful information for the patient and answer user questions with the information contained in that prospectus (via chatbot).
The system must be hosted in the cloud, where it will synchronize with CIMA via API or web scraping (weekly) saving all the necessary information (text, data, images, links to videos, etc.) only for authorized, marketed, and non-hospital use medications. The mobile application will connect to the cloud and make the necessary requests.
he recognition of the medication with the camera can be done with OCR and/or ML techniques. The user must also be given the possibility to identify the medication with a form requesting the name or National Code.
The summary and answers must also use an ML model according to the project's needs
The objective of this project is to develop a web application that allows users to restore old photographs in a simple and effective way. This platform will integrate various advanced image processing techniques and artificial intelligence to carry out the restoration. Users will be able to choose from different options to customize the restoration process according to their specific needs.
Restoration can comprise different subtasks, such as restoring the original color (or at least more vivid color tones) or filling in parts of the artwork that are deteriorated or have been lost.
You can create your own models, use open models such as Stable Diffusion or apply classical processing techniques with libraries such as OpenCV.
With the rise of renewable energies based on the solar field, the research and application of an archaic system such as the heliostat is proposed, applying it to new technologies that can automate and optimize the angles of incidence and redirection at certain times of the solar cycle to the photovoltaic receiver cells.
As the angles of installation of solar panels are usually limited in terms of positioning and working hours of direct incidence, the aim is to study and create a small-scale prototype of the device responsible for redirecting through refractors (mirrors or other reflective surfaces) at times when their efficiency suffers a considerable decrease and its redirection from the prototype is feasible.
In the first part of the project, the prototype will be developed on a feasible scale using 3D printing technologies, servomotors and an Arduino controller to locate the most intense point of light and its optimal degree of incidence. Subsequently, a model of reflective surfaces will be used to redirect the light to a desired point chosen by the user.
As a final section, it is proposed to make an application study in real environments towards photovoltaic receiver cells, along with a calculation of performance improvement with a full-scale prototype to be applied with the data obtained in the previous stages.
For the realization of this project, Arduino modules, servomotors and light sensors will be provided to the student. The 3D printing can be carried out by the student himself/herself if he/she has the equipment or, otherwise, it can be carried out by the main tutor who does have such infrastructure.
In today's society it is very important to know whether the message being transmitted produces the desired reaction in the target audience. Identifying whether a sad scene in a movie or a part of a political speech produces the desired effect on a control sample can translate into achieving the desired objectives when exposed in public.
The goal of this project is to use facial recognition technologies to detect live audience expressions and to know if the message is producing the desired reaction in the audience.
It will be necessary to identify how many people, based on images, are watching the show in question. Subsequently, classification models will be applied to detect the type of feeling expressed by each of the spectators, whether it is joy, sadness, laughter, interest, boredom, anger, etc. Publicly available models may be used, or one's own models may be developed, as well as finetuning of third-party models. The student will be responsible for finding or generating the appropriate datasets for this task.
Noise pollution becomes a more severe problem every year, affecting the mental and auditory health of the population. One of the most characteristic sources is construction works. Moreover, construction works are problematic because they need to be carried out at times that don't disrupt traffic and are restricted to a specific time frame for making noise.
The aim of the project is to create a prototype to help with this issue. The idea for the final product would be a kind of tripod placed around the construction site. When the machinery is turned on, the device would significantly reduce the noise through active noise cancellation (ANC). This would allow more flexible working hours for the construction, enabling them to avoid peak hours and the sun during its hottest moments. This device could also be used in construction projects in natural terrains, reducing the impact of noise pollution on wildlife.
The prototype would be a scaled-down version of these tripods. To perform active sound cancellation, the prototype would require microphones, speakers, an Arduino, and a Raspberry. The microphone would read the sound waves, send them to the Raspberry, and the Raspberry would send the inverse signal to the speakers to cancel the sound. The objectives would be to create a scaled prototype, test if it's truly possible to create a "sound cancellation dome" using this prototype, and analyze alternatives for this goal.
Last-minute canceled reservations (or those where diners don’t show up at all) are a serious problem for restaurants: on the one hand, they lose the use of one of their tables, with the opportunity cost this entails (along with other fixed costs such as staff, heating, etc.), and on the other hand, they may have to discard certain perishable goods they have based on that demand, such as meat, fish, seafood, etc.
On the other hand, there is also a group of people who would like to have lunch or dinner at certain restaurants with very limited availability but don’t call or visit daily to check if any reservations have been canceled by chance.
This project aims to connect restaurants and diners around canceled reservations. The goal is to develop a mobile application that fulfills a dual purpose: first, allowing restaurants to report last-minute cancellations or "no-shows," indicating at what time the table would be available again and the number of diners accepted (usually a range like "2-4 diners" depending on the table's size and the original reservation). Secondly, users can receive notifications based on their preferences and geographical location, notifying them that a table has become available, and they can confirm a new reservation through the application.
The application will therefore have two types of users: restaurants that register and offer their tables, and everyone else who subscribes to alerts about available tables. Ideally, profiles will be dual: a restaurant owner may also wish to use the application in its other mode, with owning a business being an "addition" similar to Google Business, where you add your business to your profile. However, this functionality is not 100% necessary.
It is suggested to create an Android application, either natively or through another technology (Flutter, web views, etc.), which accesses a cloud-hosted backend where the app's logic and user, restaurant, and reservation data will be managed.
Within multidisciplinary teams, it is always challenging to understand the skills and knowledge each member possesses. These skills tend to be individualized rather than treated collectively as a team resource.
For this reason, the development of a web application is proposed to help identify the current skills of a team, analyze them, and generate metrics on their distribution. The application will enable the creation of graphs and reports for further analysis, allowing users to assess potential risks within the team.
Examples:
Too many team members with certain skills and too few with necessary skills.
Analyzing team members' adaptability to new challenges that require new skills.
Training and certification planning for the team.
The project will consist of at least three main components:
Back-end:
A cross-platform back-end service (Linux and Windows) with a REST API implementation will be required.
The choice of programming language and REST API framework will be up to the student.
Front-end:
A front-end interface will be needed to manage and visualize the application.
The choice of language is up to the student, keeping in mind that it must be designed to display multiple metrics through graphs and reports.
Database:
A database must be created with tables based on a non-relational model for data storage.
A relational database may be used for managing application-related data.
The Montemadrid Foundation is a non-profit organization that works towards inclusion, access to education, employment, culture, and environmental conservation. As part of its programs, the foundation works with children with ASD (Autism Spectrum Disorder) and helps them face everyday situations that they do not understand and in which they do not know how to act. To achieve this, they explain to them in a personalized way what is going to happen, what they can and cannot do, where they can do it, etc., in those moments that are so confusing for them.
This project involves developing a platform for creating social stories, presenting them in an entirely visual way and introducing at the beginning of each story where the situation takes place, when it happens, what will happen first and what will happen next, and with whom it will occur. Subsequently, the story will unfold, and the child will interact with the platform, evaluating whether they act appropriately, always with the support of trainers and tutors.
It would be desirable for the situations to be customizable with photos of the children and their families, as well as real places where the events take place, such as their home, school, a neighborhood store, etc. This will eliminate abstraction difficulties and ensure the comprehension of the story. For this, artificial intelligence techniques for face replacement or more traditional methods such as drawings, masks, and others can be used.
Regarding implementation, it is desired that the platform for generating stories be a web platform or a mobile application. Additionally, it should feature an interface for administrators, i.e., members of the foundation, allowing them to configure each story and customize them for each child. Therefore, the stories will also be associated with a specific user, which implies adding a user management and permissions system.
Infrastructure as code has become a standard in the process of cloud infrastructure deployment. Terraform, tool widely used in this sector, has recently modified its license, prohibiting some uses allowed before.
On the other hand, Nix, a declarative language for describing system’s configuration, is a tool growing in popularity in dev fields and by DevOps profiles, given the expressiveness of the language and its capacity to define reproducible environments.
The project aims to build a tool capable of analyzing the differences between the selected platform (Kubernetes, AWS, Azure..), and the generated configuration from Nix files and be capable of performing the necessary reconciliation operations.
The system should be modular by accepting the import of plugins alike terraform accepts ‘providers’, so it wouldn’t be necessary to define all operations for all cloud providers in the market but allow the definition of new ones.
Nowadays, social media entertains us as users, but for many brands, it is their main promotional tool, where they gain visibility or simply seek to enhance their brand image positively. However, the path is not easy; it is filled with trolls and haters who use social media as their stronghold to discredit, spread misinformation, or cancel brands.
For this reason, Troll Hunter aims to use AI to identify when a comment or review may come from a troll account or a hater, and propose to a brand a witty and creative response to improve its image.
To carry out this task, public data regarding comments on social media will be incorporated, particularly those linked to brands and their community managers with the responses they provide. Using LLMs (Large Language Models), a classification system will be generated to categorize each comment into one of the two categories (or neither), and a suitable response will be created.
This project proposes the use of Artificial Intelligence models and techniques for the generation of audio from an input video. The aim of this project is to develop a web application that allows users to add a new audio track to a video. The user will upload a video to the platform, the original audio will be removed from the video and using artificial intelligence, a new audio track will be generated synchronised with the content of the video frames.
To achieve this, it is proposed to implement a video to text conversion process and, subsequently, from text to audio, in order to generate the synchronised audio track. The implementation must consider the possible overlapping between the different sounds present in the scene, the temporal resolution necessary to achieve a good synchronisation between image and audio and whether only ambient noise is generated or the recreation of human voices.
This project is carried out in collaboration with the ASPAYM Foundation, whose mission is to promote and encourage all kinds of actions and activities aimed at improving the quality of life of people with spinal cord injuries and severe physical disabilities.
To support the recovery of individuals affected by these issues, the foundation has created specific 3D-printed puzzles designed to encourage users to perform certain physical movements: wrist rotation, elbow extension, arm abduction, etc. It has been verified that these puzzles achieve their intended purpose. However, they require individuals to physically visit the association to perform the exercises and to be accompanied and supervised by a professional from the association to analyze how the activity is being carried out.
The proposal is to create a virtual environment where individuals, through a VR headset and controllers (hand-tracking functionality could also be considered), perform puzzles analogous to the physical ones. This application will serve a dual purpose:
The project proposes using Unreal Engine, developing the logic both in C++ and BlueprintUE, and incorporating necessary assets available in the Unreal Engine marketplace, provided their licensing allows their use in this project. Lastly, the project will be carried out with guidance from ASPAYM professionals.
Machine learning requires managing big quantities of information that keeps changing over time and must be utilized concurrently by many users for different purposes, which requires a version control system.
Git has become a standard, but it is designed for managing source code (text), and the information must be cloned to access it (distributed). This means, for example, that for making a training over certain version of the dataset, it is required to download all files.
We would like to design a tool that stores, manages and version files; by using the same protocol than Git (object, commit, branch, merge, tag…), but optimized for the following cases:
