Qualysoft's AI Success Stories - Practical Applications in Production Across Diverse Industries

Reimagining industries with innovative production-ready systems that deliver real business value

 

January 23, 2024 | Read time: 6 min | Written by: Tamás BARTÓK - Qualysoft Head of Software Engineering and AI and Head of Solution Sales

Introduction

At Qualysoft, we pioneer the latest technology trends and cutting-edge innovations that shape the future. This redefines every aspect of project execution and client consultation. Our mission is to guide our clients through their digital transformation journey and to help them uncover untapped opportunities to help them become more efficient and responsive.

AI has become a popular buzzword in the IT industry, yet its significance extends far beyond a mere trend; it's a cornerstone for future sustainability and cost-efficiency. In our last article, we explored the practical readiness of AI for business implementation, technological advancements, industry applications, and provided an overall image of the Qualysoft AI services.

This new article dives into our AI project portfolio. We will share practical case studies of AI solutions that are operational in various production environments. We intend to offer real-life examples of AI, aiming to illustrate how these technologies are effectively applied in different business contexts. These case studies confirm the real-world impact of AI and its ability to enhance business processes and drive digital innovation.

Revolutionizing Restaurant Service: AI-Driven Food Detection for Enhanced Efficiency at Wasabi

Wasabi, a well-known Japanese restaurant chain in Hungary, has embarked on integrating an AI-driven Computer Vision system. The unique selling point of the restaurant is its conveyor belt system. Various dishes are presented on rotating plates, allowing clients to select and enjoy their preferred items. The primary benefit for the customer is to track the quantity and variety of dishes being picked and removed from the conveyor belt and enhance supply chain efficiency.

Qualysoft sustained Wasabi with an AI-powered computer vision application. This application is programmed in Python and utilizes Azure Cognitive Services for the segmentation and identification of different food items. It operates on Raspberry Pi hardware, to analyze and recognize the assortment of dishes presented. This enables the restaurant to gather real-time data on food consumption, facilitating informed decisions for inventory and supply chain management.

Enhancing Agricultural Monitoring: Advanced Vegetation Analysis Using Satellite Imagery and AI

Transmission lines carrying electricity must, of necessity, span large areas where crops grow. In these agricultural regions, monitoring the condition of vegetation is essential. Our system employs satellite imagery to assess the health of vegetation in these areas.

This technology enables continuous surveillance of targeted locations. It allows us to provide insights and forecasts with submeter precision.

The team composition for the project included: one Project Manager, one Senior Researcher specializing in machine vision, two Senior Data Scientists, one to two Junior Data Scientists, one Senior Developer, two to three Mid-Level Developers, two Technical Staff members, and a Sensor Development Team consisting of one to four individuals.

Technical details- AI Focus in Computer Vision and Deep Learning with Neural Networks:

  • SAM (developed by Meta), a state-of-the-art model for advanced image processing.
  • YOLO (You Only Look Once), a real-time object detection system.
  • CLIP (Contrastive Language–Image Pretraining) by OpenAI, designed for understanding a wide range of visual concepts.
  • Detectron, a versatile framework for object detection and segmentation.
  • MASK-RCNN, a powerful model for object instance segmentation.
  • Transformers, such as DETR (Detection Transformer), for innovative approaches in image recognition.
  • Auto-encoders, used for efficient image encoding and reconstruction tasks.

The project's development leveraged Microsoft Azure Cloud, encompassing both Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). Additionally, integration with SAP Hana was executed on the client's end.

Smart Meter Reading Revolution: Automating Accuracy with AI-Enhanced Vision Technology

Electric meter readings can currently be captured using basic meter reading apps that transmit photos or through manual keyboard entry. However, both approaches are prone to errors. Manual entry is susceptible to typos, and photos sent through the app may be blurry or poorly lit, making data difficult to decipher. Additionally, in the case of photo submissions, meter readings often require processing by customer service staff.

To address these issues, our application provided the energy service provider with a machine vision module. This feature eliminates the need for manual intervention by automatically detecting the serial number and scale value of the consumption meter using the smartphone's camera. The data is accurate and instantly logged into the invoicing system.

This approach not only saves time and eliminates redundancy tasks, but also reduces costs, as it eliminates the need for meter readers or customer service involvement. The entire reading and recording process becomes fully automated.

 

Project Team Composition: The team encompassed one Project Manager, one Senior Researcher with expertise in machine vision, one Senior Data Scientist, and one to two Junior Data Scientists.

Development Platform: The project development leveraged Microsoft Azure Cloud, including both Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). Additionally, SAP integration was performed at the client's site.

Application Integration: The developed application has been integrated into customers' customer service mobile applications.

Support Agreement: Under our support agreement, we provide solutions for server-side evaluation module issues and continuously update our models to adapt to new meter types.

Next-Generation Quality Control: AI-Driven Audio-Visual Inspection for Enhanced Manufacturing Safety and Efficiency

Our automated quality inspection system leverages audio-visual data collection and machine vision to accurately classify products into predefined quality classes. This system not only identifies potential hazards in real-time through audio and video monitoring, ensuring a safer workplace but also anticipates failures and events, enabling proactive measures to prevent work-related accidents.

This solution significantly reduces the production of defective products, leading to enhanced cost efficiency and improved quality assurance. In addition to AI capabilities, we offer the integration of physical robotic arms, developed in collaboration with our partners, to further streamline and optimize the manufacturing process.

 

Project Team Composition: The team encompassed one Project Manager, one Senior Researcher specializing in machine vision, one Senior Data Scientist, two Junior Data Scientists, and one Senior Developer.

Solution Architecture: The project utilizes a sophisticated on-premises setup, comprising custom-made sensors linked to advanced server-edge hardware. This dedicated hardware serves as the nerve center for our system, hosting both the evaluation and display modules. We have meticulously designed the architecture to ensure minimal integration with client-side systems, thereby streamlining the implementation and reducing potential complexities.

Revolutionizing Customer Service: AI-Desk's Automated Call Center with Advanced NLP and AI Integration

This innovative project was collaboratively developed with a market leader and an expert in call center solutions. The final product is a comprehensive GPT-based system, designed to revolutionize call center and technical support operations.

It seamlessly integrates across all primary communication channels — chat, phone, and email — ensuring precise and timely responses to customer queries. A distinctive feature of this system is its 'co-pilot mode', where AI and human operators work together, combining the efficiency of AI with the deep human interaction understanding to deliver superior customer assistance.

 

Project Team Composition: The team encompassed one Project Manager, one Senior Researcher, two Senior Data Scientists, one Senior Developer, one Mid-Level Developer, and one Senior Frontend Developer.

Technical details - AI Focus in Natural Language Processing and Multimodal Generative Models:

Vector Embedding and Semantic Search Databases:

Tools for AI and NLP Development:

Technology Platform: The project leveraged Microsoft Azure Cloud, leveraging Infrastructure as a Service (IaaS), Software as a Service (SaaS), and Platform as a Service (PaaS), with a focus on language models using OPENAI services.

Pilot Projects: We have launched pilot projects with multiple clients, each tailored to address specific needs. These initiatives encompass comprehensive support ranging from infrastructure setup to troubleshooting and resolving errors. Additionally, we focus on the continuous development and refinement of models, ensuring they align with the evolving requirements of our clients.

Qualysoft - Your AI Development Partner

AI has stepped out of the realm of science fiction, and now we’re seeing it practically every day, across every industry. From agriculture to call centers, these real-world applications of AI are shaping our present and redefining our future.

87% of organizations believe AI and machine learning will help them grow revenue, boost operational efficiency, and improve customer experiences, according to research firm Frost & Sullivan's Global State of AI Report.

We infuse AI capabilities in modern business applications to help you automate repetitive tasks and expedite decision-making. We are committed to helping our clients meet their business goals by offering cost-effective, result-oriented, world-class AI development services.