AI Solutions Provided by BA-AI Lab
Highlighted Research Projects
Semantic Interior Mapology
We introduce the Semantic Interior Mapology (SIM) toolbox for the conversion of a floor plan and its room contents to a vectorized form. The toolbox is composed of the Map Conversion toolkit and the Map Population toolkit. The Map Conversion toolkit allows one to quickly trace the layout of a floor plan, and to generate a GeoJSON file that can be rendered in 3D using web applications. The Map Population toolkit takes the 3D scan of an interior and populates individual objects of interest with an accurate dimension and position within the GeoJSON representation of that building.
HLR-based Model for Detecting Fake Reviews
A hierarchical logistic regression (HLR)-based model for detecting fake reviews that considers both linguistic and behavioral characteristics. With this outcome, our kernel also has multiple applications, including the detection of review spammers as a pre-module of quality in machine learning. The experimental results demonstrate that HLR can classify fake reviews and review spammers more accurately than the standard machine-learning algorithms.
Local Graph Point Attention Network
A novel end-to-end trainable graph attention network that extracts global features in terms of local graphs. Our network presents a general local graph which obtains the most fundamental features based on point order positions in different neighborhoods. Central point attention is introduced to share weights with neighboring points to reinforce central point impacts.
Mobile Robot: Speech Recognition for Automation and STEM Education
A technical model to design a Mobile Robot that combines user interaction screen, voice recognition, and AIDL IPC interactive model for remote control. Our robot framework employs RockChip AI Processor RK3399Pro, IPC AIDL architecture, and Automatic Voice Recognition technology to maneuver a robot in three languages: Korean, English and Vietnamese.
Model for Collecting, Analyzing, and Identifying Trends based on Customers Feedback
This work proposes a model for collecting and analyzing data based on customer feedback in two-folds: (1) Building a model for data collection from multiple resources; and (2) Experimenting on the latent topic mining and identifying customer discussion trends.
Personalized Recommendation Model in E-commerce
A comprehensive approach to engage users on e-commerce platforms through an implicit personalized product recommendation engine. This research combines the strength of several recommending algorithms, consisting of collaborative filtering, popularity, and Bayesian personalized ranking, to develop a robust recommendation system. By leveraging a retrieval strategy and evaluating candidates through a series of ML models, our system addresses the challenges of analyzing large-scale data, cold-start problems, and personalization; enhancing user experience and driving sales.
Industry-sponsored Projects
Onsite Safeguard System at Hoa Sen Group
AI-based application to improve human resource management, timekeeping, and workplace compliance. The system is integrated with ERP (Oracle) to provide seamless data flow and optimal operations across the production process.
Digital Transformation of Smart Agricultural Management
This project aims to digitally transform the cultivation and management of farms in central provinces of Vietnam. The scope of work includes managing the cultivation process, monitoring environmental conditions, and automating irrigation and fertilization for greenhouses.
Autonomous Coffee Bean Quality Assurance
AI-based application to (1) identify and estimate the percentage of non-quality coffee beans, (2) forecast future prices, and (3) build strategies for buying, selling, and hedging coffee beans on markets.
Smart Traffic Surveillance System
The goal is to improve public safety by integrating AI-based tracking technologies into the current camera grid. Key functionalities include vehicle-type detection, license plate recognition, and individual identification. Incorporating with location of cameras on grid, vehicle detection is traceable on Google map in real-time.