AI Research and Development

With an expertise team in Artificial Intelligence (AI), one of our strong points is the abilities to help customer consulting and solving their daily business operating issues by applying AI state-of-the-art technologies. We can develop full-stack AI software solution, which covers the tasks from business insight, data mining and analysis, model development, model deployment on both edge and cloud, etc., in order to provide the ultimate solution for a variety of important use-cases.

Computer Vision 01
Natural Language Processing 02

03 Speech Signal Processing
04 Business Data Analysis

Case Study

Technological and Human-Oriented Research Lab

Technological and Human-Oriented (THOR) Lab is a new applied AI research and development lab in the central of Vietnam, Danang city. Founded by D-SOFT JSC in mid 2020, our vision is to create a modern R&D lab, where the state-of-the-art technologies meet the needs of software solution industry. We mainly focus on conducting high-quality research while integrating AI in software development not only for D-Soft productions but also for our customer requirements.

AI Camera System to analyze customer behavior

This project is to develop an AI Camera System PoC that can support analyzing customer behavior in supermarket.

The system includes 3 modules:

+ Webapp on Cloud side to receive detected results from AI Camera and visualize on Dashboard.

+ AI engines are embedded on AI Camera to detect customer on-edge, this application has been integrated with Person Detection, Pose Estimation model for customers and their behaviors detection.

+ Mobile app to receive alert notifications from Webapp any time the system detects suspicious behaviors for security purposes such as crime prevention.

Client	Japan	Technologies	VueJS (Front-end), PHP Laravel (Back-end), Python, AI, Mobile (iOS, Android)
Domain	Service	Scope	18 man-month (6 man x 3 months)

Offline shopping behavior analysis

This project is to develop an PoC application that can support analyzing customer behavior in supermarket offline via recorded video files.

This can help the owner understand which products are focused/touched by customers daily and take necessary actions to improve store layout for productivity and customer satisfaction.

The application includes 1 core engine that can import multiple video files simultaneously, detect customers & their hand coordinates in the videos and output to CSV files touched/not touched status on any products.

Besides, the application can also detect whether a detected person is a customer or shop worker based on Worker/Customer Classification engine.

Client	Japan	Technologies	Python, AI
Domain	Service	Scope	10 man-month (5 man x 2 months)

Offline train ticket usage tracking

This project is to develop an PoC application that can support analyzing customer behavior at the station (for ticket usage tracking, passenger counting) offline via recorded video files.

The application includes 1 core engine that can import multiple video files simultaneously and support other features below:

+ checking ticket usage for all passengers passing the station gate

+ counting the passengers entering/exiting the station

+ tracking passengers in dangerous nearby-the-train area to alarm to station guard (in urgent)

Client	Japan	Technologies	Python, AI
Domain	Service	Scope	15 man-month (5 man x 3 months)

Offline food store monitoring

This project is to develop an PoC application that can support analyzing customer behavior in mini restaurants offline via recorded video files.

This can help the owner monitor the status of every seat in their stores automatically (empty, ordering, waiting, eating…) to provide services accordingly as fast as possible