Automated Quality Control with AI in Cement Industry​

Installation on Cement Production Line

Automated Quality Control with AI in Cement Industry

A Vision Solution by Irida Labs and Basler

Installation on Cement Production Line
In collaboration with

Cement Industry Production: A Fast-Pacing, Harsh Environment

Cement industry provides an example of heavy industrial manufacturing process, involving multiple, complex stages before the end-product gets ready for shipping. A typical cement plant has a production capacity of several thousand tons per day, which is translated to a continues output of tens (or hundreds) of bags per minute on a 24/7 basis.  

In such a fast-pacing production line, every halt (i.e. to remove a defective bag) is translated to a significant delay in the production, packaging and loading flow, which in turn impacts the overall efficiency and productivity KPIs. On the other hand, every undetected defective cement bag risks contaminating an entire pallet, not to mention hurting client relations and adding an excessive cost for returns, insurance or reimbursement. On top of the above, due to cement’s nature as a chemical product, legislation dictates very specific labelling requirements related to chemical, environmental and other specifications; failing to comply (due to printing or other failure) can lead to legal or financial consequences. 

QA/QC Challenges for the Cement Industry

In order meet top-quality standards, this leading European cement manufacturer utilized human inspector shifts to perform a 24/7 in-line visual inspection on the produced cement bags.  

The purpose of the project was to introduce a real-time quality inspection solution on cement bags that could identify any defects in random locations on the product’s surface, such as cracks, spills, dents, cuts, damaged edges or even printing failures. The solution should provide always-on defect detection on continuously running conveyor belts, while providing real-time alerts when a defective product is detected. Integration with (pre-existing or new) 3rd party systems should also be available, in order to handle these alerts and trigger specific actions (i.e. alarm, PCB, automated disposal etc), not to mention the collection of production analytics and statistics. 

Last, but not least, the system should be robust against changing lighting conditions, excessive dust & dirt, vibrations, while being able to adjust to various packaging types (cement bags of different sizes and colours). 

Damaged Cement Bags
Damaged Cement Bags
Packaging Integrity & QA cover
Human Quality Control on Cement Production Line.
Image Credits: © picture-alliance/dpa/U. Anspach

Human Visual Inspection

At its previous state, the packaged products Quality Control and Quality Assurance was performed via human visual inspection. This used to be a labour-intense process, involving 3 shifts of experienced operators to continuously oversee the production output, manually pause the conveyor systems in case of a detected defect, manually remove the defective bag and restore the operation. This process was highly prone to human errors, human fatigue even optical illusions, not to mention the sub-optimal efficiency and output due to the inevitable delays. Trustworthy data collection regarding the defects was also insufficient, making QA/QC process non-measurable and thus harder to improve. 

Artificial Intelligence is the game changer in this area, introducing next generation Industry 4.0 Vision AIoT solutions ,  allowing  numerous Smart Factory applications. Specifically on the QA/QC, Vision AI surpasses human vision in quality and quantity measurements because of its speed, accuracy, repeatability and scalability. Combined with the necessary optics, Vision AI can also detect defects that are impossible to consistently spot with the naked eye! 

The Computer Vision & AI Packaging Integrity & QA Solution

The implemented Computer Vision & AI Packaging Integrity & QA solution is based on and offers full automation in the Quality Assurance and Quality Control processes.

On the ground, each camera sensor for the Edge Vision AI Solution is paired with an on-premises processing device that operates as near as possible to the sensor. Each sensor is positioned at the optimal distance/height/angle indicated by Vision System Design, so the products are within its field of view and the characteristics-to-be-monitored are visible (aka 5 out of 6 sides of each product). The Vision AI solution is comprised of the following components: Vision AI Software & Services Platform

The solution is powered by, Irida Labs’ end-to-end software & services platform for deploying Vision AI at scale. handles the heavy lifting of orchestrating all the building blocks required to deploy a Vision AI solution; Vision Sensors & Edge Devices, Digital Vision Twin, Edge Hardware, Vision AI Software, Data Engine, APIs and AI Analytics are all integrated into the platform, streamlining development to support of both POC as well as full-scale production.

Components of the Vision AI Solution

Fully custom Vision AI software, tailored to each plant production specifics. Under the hood, platform for on-device Vision Intelligence is utilized to build efficient and robust Defect Detection, Packaging Integrity and Quality Assurance solutions, that run in real-time and rely solely on our proprietary ML engine. With the use of state-of-the-Art Vision AI algorithms, it is capable of identifying spills, dents, cuts, damaged edges, even printing failures with over 99.9% accuracy, while continuously learning and adjusting on new types of defects and products.

The host device is an Edge Device or Industrial PC utilizing an AI accelerator (such as ADLINK NVIDIA® Jetson Nano™ Edge Inference Platform), being able to provide the sensor with the computational power required to run advanced AI vision workloads.

Dashboard & AI Analytics are powered by providing real-time visual check of the functionality of the system. All metadata can also be transmitted to pre-existing software stack (i.e. ERP, WMS etc.) via the industry standard MQTT or HTTP protocols.

Basler ace series cameras (2 USB color cameras for defect detection and 1 Gige monochrome camera for printing failure checks) along with appropriate lenses are mounted on rigid brackets over the production line, triggered by a photocell for capturing the cement bags at the same time, therefore being able to inspect all bag sides on the same conveyor belt spot.

Camera module and optics are protected from the industrial environmental conditions (vibrations, dust, moisture) with a compact IP66/IP67 camera enclosure.

Vision AI packaging integrity solution
Defect detection and production analytics in action on Platform

Get in Touch with our Team!

Explore further how our Vision AI Solution for Packaging Integrity & QA, powered by, can help automate the integrity inspection and quality assurance procedures at your company.

We would love you to give us a very brief description of your implementation scenario, so that we assemble the most qualified team of Vision AI experts for our discussions!

    Technical Specifications Of Hardware And Software
    Camera System
    Top-down / angled
    Real-time & HW-times stream
    > 99.9% System Level Performance
    Image Resolution
    2680 x 1529
    Indoor / Outdoor Data Analytics service control dashboard API for 3rd party integration
    Real-time video/image recording & alerts

    Benefit of Using Vision AI in the Defect Detection System

    It’s been over one year since the Vision AI Packaging Integrity & QA solution has been operational at the production line of TITAN. During this time, the company reports some remarkable results:  

    Zero-Defect Palletizing

    The 24/7 automated unsupervised alerting on defects that require immediate attention/action constitute an immediate added value to the company, contributing to 95% reduction of defective bags being palletized as well as reduced Cost of Poor Quality (COPQ). Operations managers have a full overview of entire production process, in a single or even multiple sites, and they can react in real-time if/when needed. 

    Zero Returns

    As an immediate consequence, the solution contributes to Zero Returns, since no defective products get palletized and get shipped to clients.  

    Labour Cost Reduction

    80% Reduction in Labour Cost, since the solution eliminates the need for a manual human inspection of cement bags.


    Compliance with local / EU / International Regulations (label/date misprint), reduced regulatory fines risk 

    Efficiency in Production Processes

    The system provides a real-time output when a defect is detected, which can in-turn trigger rejection systems, alarms or other automations. Detailed production analytics are being collected at the same time, providing an overview of the production output. All critical KPIs of the cement manufacturing process, including cement bag counting, defects detection (spills, dents, cut or damaged edges), printing failures (missing stamps, production date), are gathered in a single dashboard, allowing the real-time overview of one or multiple production lines from anywhere in the plant. 

    • Automatic, concurrent image capturing based on camera triggering mechanism 
    • Capture data / images during system operation and store to the local Industrial PC 
    • Visually check the Defect (or Production Date) that the AI system detected in real-time 
    • Data analytics and graph-based statistics (plots) 
    The solution significantly reduced the number of defective sacks being palletized, with accompanying reductions in labour costs. As a consequence, customer complaints about defective sacks have significantly decreased
    Grigorios Maravas
    Packaging Manager at Titan Cement Company SA at Patras Plant
    By using high-quality Basler cameras and our state-of-the-art AI vision software powered by our “” platform, we enable a system to understand the visual information of a camera like a human eye, and turn it into useful data and insights for further processing and utilization. In this way we provide a high-fidelity and highly accurate system for packaging integrity and quality assurance inspection.
    Christos Theocharatos
    COO of Irida Labs and Project Manager of the AI vision solution

    Get in Touch

    Explore how our Vision AI solution can help automate the integrity inspection and quality assurance procedures at your company. 

    Image Gallery

    Industry Brief

    In this Cement Industry Brief we present how Irida Labs’ platform for on-device Vision Intelligence is utilized to build efficient, robust and scalable Industry 4.0 solutions for Cement Bags Identification & QA in conveyor belt systems that run in real time and rely solely on our proprietary ML engine. 

    Comments are closed.
    Skip to content