Category: Subdegree

Categories
Subdegree ypec-2024

SD08 – AudioSense

he AudioSense project aims to improve the accessibility and understanding of visual content for visually impaired individuals in today’s multimedia-rich era. By leveraging innovative artificial intelligence technology, our tool can automatically identify video content, generate accurate descriptive text, and convert it into auditory information through speech synthesis. This enables visually impaired individuals to “shape by sound,” enhancing their ability to access digital information and improving their quality of life. AudioSense surpasses the limitations of traditional audio description production by increasing efficiency and providing visually impaired users with a more immersive video experience through high-quality descriptions. Additionally, our system supports multiple languages, making it globally accessible to visually impaired individuals regardless of language or region. The innovation of this project extends beyond technological advancements, as it contributes to enhancing social inclusiveness and promoting barrier-free access to information. The AudioSense development team is dedicated to creating a more open and equal information society for everyone through continuous technological innovation.

Categories
Subdegree ypec-2024

SD11 – Project: School AI

The HKDSE examination places significant milestone on both students and teachers during their secondary school life, with teachers striving to maintain a strong reputation for delivering exceptional results, while students seek favorable outcomes to facilitate their academic progression. In response to these challenges, our project aims to develop an innovative system that supports intelligent learning and caters to the diverse needs of students. The proposed system will serve as a reliable and accessible platform, capable of handling the teaching load and operating 24/7. Students will have the flexibility to ask questions and seek clarifications about their syllabus knowledge at any time, facilitating a continuous learning experience. Additionally, the system will employ intelligent algorithms to simulate public examination papers, enabling students to practice and evaluate their performance. Through real-time feedback and personalized recommendations, the system will assist students in identifying areas for improvement, enhancing their overall exam readiness.

Categories
Subdegree ypec-2024

SD12 – Language Navigation Intelligence for Robot

In recent years, products and software related to generative artificial intelligence have seen rapid development. Examples include ChatGPT and Midjourney. These large language models offer people significant convenience and access to a wealth of rich content, with very low threshold requirements. Furthermore, we found that large language models can act as a bridge between robots and humans that allow humans to command robots by language, and robots able to understand our verbal command correspondingly. This finding brings innovative ideas to robot remoting aspects. This project seeks an innovative approach to robotic remoting. We found that there are a series of issues with typical robotic controlling approaches like the long learning path to the robot controller, and the limited dexterity and precision of the human. Therefore, We tried to develop a robot-understandable translator to execute our command. To achieve this, We developed Integrate Advanced Natural Language Platform to translate our verbal language into robotic commands. Also, we developed a Seamless Voice-to-Text Conversion and communication platform to allow the controller to control the robot by month or texting respectively. We believe that this innovative approach can reduce the learning cost of robotic remoting and also be the most effective and the simplest way to control different types of robots.

Categories
Subdegree ypec-2024

SD14 – Rat Colony Detector

Rat infestation has been a long-standing problem for humans, especially in high-density cities such as Hong Kong. In addition, the structure of vertical cities makes the problem more complicated. One of the main reasons is that rats can obtain food from restaurants, but humans lack ways to eradicate them. Eventually, the mice continue to multiply and maintain a certain number. The purpose of this project (Rat Colony Detector) is to provide an effective way to eliminate rats. By installing a GPS and altimeter on the mouse, the mouse’s path can be tracked. We also collected data on the gathering, work and rest of rats, and finally estimated and discovered the nest. At the same time, through visual images, it provides citizens and rodent exterminators with visibility and understanding of the severity of rodent infestation in the area. This move is expected to narrow the scope of rodent control, reduce the cost of rodent control, and greatly improve the effectiveness of rodent control.

Categories
Subdegree ypec-2024

SD10 – PURE BLUR

In the age of information technology, platforms like YouTube have increased, allowing people to upload videos and watch videos and streams. Despite efforts to enforce censorship, inappropriate content such as sexual and violent still proliferates on these platforms, negatively impacting teenagers who may imitate such content. Furthermore, YouTubers and streamers have become popular professions, but issues like privacy rights and legal troubles arise when recording someone’s face without consent. The demand for applications addressing these challenges is increasing. PURE BLUR is introduced as a video-blurring application with live-streaming capabilities to tackle these issues. It employs algorithms and machine learning models to automatically detect and blur sensitive areas or objects in videos, ensuring that inappropriate or explicit content is minimized. PURE BLUR also encourages users to maintain a safe video-sharing environment by reporting inappropriate content. This feedback enhances content moderation and improves the accuracy of the blurring feature. In conclusion, PURE BLUR presents a comprehensive solution to the challenges faced by media platforms. By leveraging advanced technologies, user feedback, and active moderation, it aims to create a safer environment, reducing the sharing of inappropriate videos and protecting individual privacy.

Categories
Subdegree ypec-2024

SD09 – AI-Driven Digital Twins: Smart IoT Centre Management System

The project titled ‘AI-Driven Digital Twins: Smart IoT Centre Management System’ aims to bridge the gap between the physical and virtual realms by utilizing digital twin technology. It involves the creation of a 3D model of a VTC Smart City Innovation Centre, which incorporates real-world sensors and smart devices. Centre Administrators can remotely control these devices through the implementation of an AI prompt functionality, using simple prompts. The project primarily targets property management professionals responsible for overseeing IoT facilities, providing them with effective tools for system management, real-time data monitoring, and resource optimization. The project addresses challenges associated with real-time monitoring, efficient resource allocation and predictive maintenance. By seamlessly integrating IoT devices, the system offers valuable insights for informed decision-making. The project’s distinctive features include a simple AI Prompt interface, a visually appealing 3D model, and a multidisciplinary team with expertise in Electrical Engineering and Computer & Electronic Engineering. Short-term objectives involve developing a prototype and expanding the range of supported devices, while long-term goals focus on refinement and scalability. Ultimately, this project empowers administrators to enhance operational efficiency, foster collaboration, and make well-informed decisions, thereby contributing to the development of smarter and interconnected cities.

Categories
Subdegree ypec-2024

SD06 – Gemini Pro Vision AI Screen Reader

Generative artificial intelligence (Gen AI) such as Google Gemini can be used to bridge the information gap of internet access for visual impairments. The problem being addressed is the inability of visually impaired individuals to access image information due to the lack of adherence to W3C web accessibility initiatives by websites. Currently, about 60% of websites lack meaningful alternate text for their images. Moreover, it is unfeasible to retroactively add descriptive text to all existing websites manually. Sometimes, the provided web image description may not fit what visual impairments want. As a result, we’ve turbocharged the traditional Google ChromeVox Classic Screen Reader with the mighty power of Google Gemini Pro Vision to tackle the challenge of web image information access for those with visual impairments. Not just providing automatic descriptions for internet images, but also as an assistant for visual impairments daily use. For instance, when they are shopping online, the Gen AI provides detailed descriptions on the commodities such as the appearance and the usage. It helps them to buy stuffs online confidently. By utilising the popular and fully functional opensource screen reader, we can quickly provide a production-ready and affordable solution to the visual impaired people.

Categories
Subdegree ypec-2024

SD07 – SafeBuild Guardians: AI-Powered Drone & RoboDog Safety System

The goal of the SafeBuild Guardians project is to ensure construction worker safety through the integration of AI-powered drones and robotic dogs. The robotic dog system uses Google Cloud Platform and AWS, incorporating services such as Gemini Pro Vision for pipe crack detection to enable preventive maintenance. It also utilizes AWS PPE service to ensure PPE compliance. Additionally, the robotic dog is equipped with gas sensors for detecting toxic gases like methane, chlorine, and alcohol, enhancing site safety. It features a speaker and microphone to enable direct communication with workers, facilitating immediate responses during emergencies. The AI-powered drone also employs AWS PPE service to monitor PPE compliance from an aerial perspective, covering larger areas efficiently. By combining these technologies, the system provides comprehensive site coverage, offering 360-degree monitoring with no blind spots. This dual approach enhances preventive measures by detecting issues early, significantly reducing accident risks. The ability to communicate directly with workers through the robotic dog ensures rapid response and coordination during critical situations. SafeBuild Guardians leverages cutting-edge technology to create a safer construction environment, protecting workers and improving operational safety standards, making it a vital innovation in the construction industry.

Categories
Subdegree ypec-2024

SD13 – Safety Detection For The Work Of Heights

In the first half of 2023, the number of occupational injury cases in our region reached a staggering 14,197, with a fatality rate of 9.7 per thousand employees. Industrial accidents, particularly within the construction sector, accounted for a significant portion of these incidents. In response to this alarming trend, our project aims to develop a Smart Site Safety System that aligns with the government’s initiative for enhanced safety measures in construction sites.. Our project focuses on developing a model to assess the safety of three main industrial tools: ladder platforms, hop-up platforms, and mobile lift tables. By leveraging machine learning or deep learning techniques, we aim to develop a monitoring intelligence that can accurately detect and evaluate the safety aspects of these tools. The detection results will be showcased on a streaming platform, providing users with immediate feedback, and promoting the correct usage of industrial equipment. Therefore, It can enhance workplace safety and foster a culture of responsible tool usage in industrial settings.

Categories
Subdegree ypec-2024

SD01 – SoundScene – Mobile Auditory Orientation Training System

The project introduces a novel Auditory Orientation Training System (AOTS) named SoundScene, designed to enhance spatial awareness and independent navigation for visually impaired individuals. This system, leveraging AI technologies and mobile platforms, consists of two interconnected applications—one for trainees and one for trainers—operating over a local WiFi network using peer-to-peer connectivity. The trainee application uses spatial audio and augmented reality (AR) to create a 3D virtual environment, enabling users to orient themselves using auditory cues. This is achieved through the device’s camera and motion sensors, which adjust the spatial audio output to simulate real-world auditory scenarios. The trainer application facilitates real-time monitoring and dynamic scenario adjustment, enhancing the training experience. SoundScene can function in both connected and disconnected modes, ensuring versatile use in various settings. Evaluated through qualitative and quantitative measures, the system demonstrated significant improvements in the orientation and mobility skills of visually impaired users. It offers a scalable, cost-effective solution that harnesses widely available mobile technology, opening new avenues for training and assistance for the visually impaired, while also benefiting professions requiring enhanced auditory orientation.