Deep Learning for Personalised Remarketing
Lead Research Organisation:
Brunel University London
Department Name: Computer Science
Abstract
The concept of the neural network has allowed machines to recognise and manipulate an object in a way that replicates human behaviour. Some of the fascinating achievement of the neural network includes generating deep fake videos and recognition items in an image or video which I find very intriguing. My Final Year Project was based on neural networks and I had exposure to the Convolution Neural Network (CNN) and learned how to apply that in my code. This helped my project to recognise limbs on human on video footage and made it possible to describe the actions that are being carried out.
I have got proficient experiences in two programming languages which are Java and Python; most of my programming coursework has been completed in Java at my time at Brunel University and I have self-taught Python when I was in year 11 as a hobby. Learning PythonThis hobby helped me to complete my final year project and achieve a fantastic grade. Working with a start-up company (ReWallet) has further strengthened my programming skills in both Java and Python and that would be beneficial when working on the upcoming PhD project.
Working with teams is my strength as it is evident from my grades in the Year 2 Group Project. Firstly, being friendly to others and frequently communicating with other team members in my Year 2 Group Project has made coding more enjoyable and it kept me motivated. This resulted in producing a functional android app that we as a group were happy of what we accomplished. Secondly, the grade that I achieved in my Final Year Project shows that I am also confident in working independently and able to produce high-quality results. I can concentrate for a long time when I work independently thus it allows me to solve complex problems very quickly. My Final Year Project had required me to endure a significant amount of stress and solve as coding problems efficiently.
I have got proficient experiences in two programming languages which are Java and Python; most of my programming coursework has been completed in Java at my time at Brunel University and I have self-taught Python when I was in year 11 as a hobby. Learning PythonThis hobby helped me to complete my final year project and achieve a fantastic grade. Working with a start-up company (ReWallet) has further strengthened my programming skills in both Java and Python and that would be beneficial when working on the upcoming PhD project.
Working with teams is my strength as it is evident from my grades in the Year 2 Group Project. Firstly, being friendly to others and frequently communicating with other team members in my Year 2 Group Project has made coding more enjoyable and it kept me motivated. This resulted in producing a functional android app that we as a group were happy of what we accomplished. Secondly, the grade that I achieved in my Final Year Project shows that I am also confident in working independently and able to produce high-quality results. I can concentrate for a long time when I work independently thus it allows me to solve complex problems very quickly. My Final Year Project had required me to endure a significant amount of stress and solve as coding problems efficiently.
Organisations
People |
ORCID iD |
| Yongmin Li (Primary Supervisor) | |
| TASIN ISLAM (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/T518116/1 | 01/10/2020 | 30/09/2025 | |||
| 2438225 | Studentship | EP/T518116/1 | 01/10/2020 | 31/12/2023 | TASIN ISLAM |
| Description | Initially, a novel 2D Virtual Try-On (VTON) has been presented in this study, which is a deep-learning architecture for image-based clothing transfer onto a human body. Our proposed model surpasses the earlier state-of-the-art Virtual Try-On models by demonstrating superior performance in terms of occlusion and unpaired cases, i.e., converting short-to-long or vice versa. Furthermore, additional functionalities have been incorporated to enhance the appeal of Virtual Try-On for consumers. Our Virtual Try-On system has the ability to adjust the person's pose, thereby providing a wider range of viewing angles to examine how a garment may look on them. This feature further augments the overall user experience and elevates the applicability of Virtual Try-On technology in the domain of virtual apparel try-on. The primary aim of this project was to examine plausible techniques that could be employed to generate content from consumer data, such as profile pictures. The successful accomplishment of this objective was achieved through the development and refinement of these methods, resulting in their relevance for our particular use case. Furthermore, our team is currently engaged in pursuing another objective, namely the synthesis of static images into a brief video format, with the intention of arousing consumer excitement and enhancing product appeal. |
| Exploitation Route | The present project is expected to yield direct benefits to both consumers and businesses alike. Our deep learning model, which is at the core of this initiative, is anticipated to substantially enhance consumer engagement with the products and services offered by businesses. Consequently, businesses are likely to witness higher levels of consumer satisfaction, resulting in a reduction of product returns and an increase in sales. |
| Sectors | Digital/Communication/Information Technologies (including Software),Retail |