Skip to content
ProductWatch

Product Watch

Product Watch is an integrated, configurable Computer Vision and AI-based system to recognize and track products and human interactions in Industry 4.0 and Smart Retail environments.

Keywords:
Computer Vision, Artificial Intelligence, Product Monitoring
Industry 4.0, Smart Retail, AI on the Edge, Remote Management, Product Analytics

Funding Details

“Product Watch” project is carried out within the framework of the National Recovery and Resilience Plan Greece 2.0, funded by the European Union – NextGenerationEU (learn more here), with a project budget of 500,000.00 €. 

Project Description

ProductWatch is an integrated system that uses Computer Vision (CV) and Artificial Intelligence (AI) to monitor products and human interactions in Industry 4.0 and Smart Retail environments. It tracks product movements in factories and stores while analyzing consumer behavior, such as when items are picked up or placed on shelves. The system provides detailed analytics on product flow and customer engagement. Highly configurable in both hardware and software, ProductWatch leverages AI on the Edge for efficient, scalable processing. Cloud-based remote management allows easy customization and updates, offering businesses a flexible, cutting-edge tool for operational efficiency and enhanced customer experiences.

Our Role

rida Labs leads the ProductWatch project by developing the computer vision software and integrating it with customized hardware to create an advanced, configurable system. Specializing in embedded vision AI, Irida Labs ensures real-time product tracking and interaction analysis using edge-based processing for scalability and cost efficiency. The company also manages cloud-based system updates and remote configurations, supporting broad adoption. Beyond technical development, Irida Labs handles market research, intellectual property protection, and commercial promotion, ensuring that ProductWatch meets industry needs while driving innovation and efficiency across the industrial and retail sectors.

Vision

The aim of the ProductWatch project is to develop an integrated hardware + software (HW+SW) system in the form of an optical sensor that exceeds the capabilities of a simple camera. Using Computer Vision and Artificial Intelligence technologies, ProductWatch monitors the position and type of specific products as they move within a factory or store, while simultaneously drawing conclusions about human interactions with those products – such as picking items off a shelf, placing them in an oven, or moving them through a loading area.

The system produces aggregated information and statistics – on demand or on a regular basis – regarding product movement, consumer interest, and key points of activity within a controlled area.

Key Capabilities

All data processing is performed on the edge device (AI on the Edge), safeguarding operational privacy while enabling real-time response without reliance on centralized computing infrastructure.

Goals and Objectives

1.

Conduct field and market research to shape the physical and functional characteristics of ProductWatch, enhancing its innovativeness and competitiveness.

2.

Develop a complete system with specific functionality and features – including the design, implementation, and integration of all individual subsystems.

3.

Protect the intellectual property of the technologies developed, through patent applications in Greece and abroad (USPTO, EPO).

4.

Commercially exploit the system through a dedicated marketing and promotion plan, participation in international exhibitions, and targeted B2B outreach.

5.

Define Key Performance Indicators (KPIs) to optimize the system during development and guide its continuous improvement during commercial exploitation.

6.

Deliver an initial installation at a minimum of one customer – in Greece or abroad – during the project implementation period.

Technological Innovation

The development of Computer Vision solutions requires expertise spanning embedded hardware, optical system design, embedded software, machine learning, and artificial intelligence. No integrated solution currently exists on the market that combines all these elements – a gap that represents a significant opportunity.

A recent McKinsey & Company study (2020) found that 74% of companies that develop a Computer Vision pilot program ultimately fail to turn it into a fully functional product. ProductWatch directly addresses this challenge by offering comprehensive, end-to-end support across all stages: from defining system specifications and designing AI models, to managing hardware, collecting and visualizing data, and delivering the final product.

Special emphasis has been given to data privacy, as well as to the principles of Ethical AI and Explainable AI – another core point of differentiation for this solution.

Usage Scenarios

ProductWatch is designed for two primary market verticals:

Industry 4.0

Monitor and track products moving through a production line or warehouse. Detect product presence, count items, track movement, and identify anomalies - all in real time, at the edge.

Smart Retail

Analyze consumer interactions with products on shelves, at display units, or at checkout points. Understand what items attract attention, how frequently they are picked up, and where engagement is highest.

Methodology - Project Phases

Phase 1: Technology Research & Development

Detection of product presence across various scenarios (indoor/outdoor, variable lighting)

Product detection, classification, and counting in image and video streams

Product movement tracking using learning-based re-identification techniques

Detection and recognition of human actions (e.g., picking items from shelves, inserting items into ovens)

Custom object/product identification for surveillance, logistics, and retail

Development of data curation, augmentation, model validation, and ML training engines

Phase 2: Demonstration & Testing Systems

Development of demonstration systems on Hikvision DeepinView cameras (plug-in integration)

Development on Renesas RZ/V2L embedded platform

Exploration of additional platforms: NVIDIA Jetson (Xavier/Nano), Intel Movidius TPU

Development of a user interface (UI) for managing and configuring ProductWatch

Phase 3: Innovation Optimization & Commercialization Feasibility

SWOT analysis of ProductWatch across target markets (Industry 4.0, Smart Retail)

Market feasibility study covering Europe, Asia (China, Korea, Japan), and North America

Phase 4: IP Protection & Commercial Exploitation

Patent applications for deep learning technologies (USPTO, EPO, OBI)

B2B meetings and targeted outreach to expand sales networks in Europe, Asia, and the Americas

Participation in international exhibitions: VISION (Stuttgart), Mobile World Congress (Barcelona/Shanghai), Embedded Vision Summit (Santa Clara/Stuttgart), GTC Conference (NVIDIA), Electronica (Munich), Embedded World (Nuremberg), IoT Summit

Impact

ProductWatch targets two high-growth market verticals: Industry 4.0 and Smart Retail – both experiencing rapid adoption of AI and vision-based automation. By delivering an integrated, configurable, and scalable solution based on AI on the Edge, ProductWatch enables businesses to:

  • Reduce dependency on centralized computing infrastructure
  • Deploy and scale sensor networks with minimal installation costs
  • Access real-time analytics on product flow, inventory, and consumer behavior
  • Customize the solution for specific environments through cloud-based remote management

The system’s decentralized architecture allows additional sensors to be added without upgrading the computing infrastructure – making it uniquely suited to growing operations in both factory floors and retail chains. A typical example: a retail chain deploying ProductWatch across multiple locations can track product turnover, identify high-interest shelf areas, and correlate consumer engagement with sales data – without storing personal video data, fully compliant with privacy regulations.