AI-assisted real-time visual monitoring for paper production (paper industry)

Marcin Rojek
6 min readNov 25, 2020


Paper production is a multi-phase process. As per Wikipedia, modern papermaking began in the early 19th century in Europe with the development of the Fourdrinier machine (aka paper machine). This machine produces a continuous roll of paper. As it is further described on Wikipedia, a modern paper mill (=factory making paper) is divided into several sections […]. Pulp is refined and mixed in water with other additives to make a pulp slurry. The head-box of the paper machine distributes the slurry onto a moving continuous screen, water drains from the slurry (by gravity or under vacuum), the wet paper sheet goes through presses and dries, and finally rolls into large rolls. For those who are not quite familiar with the process, here is a picture of an example paper machine:

Image Credit: PIV measurements of flow through forming fabrics — Scientific Figure on ResearchGate. Available from:

One of the most energy- and capital-intensive phases is dewatering. Water removal on the paper machine is accomplished in consecutive stages i.e. press section, dryer section etc. During that phase, a natural phenomenon called “wet line”, sometimes “dry line” or generally “water line” is observed.

“Wet Line” a natural phenomenon, occurs during paper production. It needs to be carefully monitored to avoid losses and expensive breaks in production.

As described in the US patent number US4500968A, the wet line is the line of demarcation between the portion of the stock on the paper machine wire submerged in water and the portion having fibers extending above the depth of the water. Looking down at an acute angle to the wire on a paper machine as it is running one can see the glossy area where the water extends above the surface of the stock and a matte area […]where sufficient water has drained through the stock.

The wet line shall be kept within a certain area on the paper machine. Once it goes out of control, it can cause damages translating into tangible loses and delays in production. Therefore, actions need to be triggered before it gets too close to certain areas. Its presence, though, is useful to the operator for optimizing the paper machine settings. However, the operator needs to constantly inspect the machine visually, monitor the wet line manually and adjust the settings accordingly to avoid a disaster.

“Wet Line” can be monitored 24/7 with byteLAKE’s Cognitive Services. A dedicated module, designed for exactly that purpose.

To automate that process, byteLAKE has worked with the leaders in the paper industry to further optimize byteLAKE’s Cognitive Services for the purpose of automated wet line detection. The solution is visualized in the picture below.

“Wet Line” detection with Computer Vision

Our trained AI model has been designed to work in real-time and continuously analyze frames received from industrial cameras. AI algorithms inspect the surface where the fabric is being formed and trigger an alarm once a wet line is detected. In addition to that, the algorithms analyze the paper machine and the surroundings to measure and estimate the position of the wet line as well as its width. This information is then presented to the paper machine operator who can react accordingly and i.e. apply needed settings.

byteLAKE’s Cognitive Services for “Wet Line” detection has already been commercialized and successfully deployed in paper mills.

We have already successfully deployed byteLAKE’s Cognitive Services in commercial environments. Our clients leverage our AI models to streamline quality monitoring and bring AI-powered automation into various processes. The goal of these projects has mostly been to support our clients’ staff in quality monitoring efforts and eventually increase the overall effectiveness of the production processes. And in the case of the “Wet Line Detector”, to help them improve the paper production process specifically and make it more efficient.

For the paper industry, AI-powered monitoring, and measurements (i.e. distance) turned out to be the important steps that secured the whole process of paper production. A so-called “wet line”, popping up on a production line, is a normal phenomenon. However, as explained above, its behavior and development need to be carefully monitored. Thanks to byteLAKE’s Cognitive Services the process of forming and moving the wet line can be monitored 24/7.

Paper Production monitoring with Artificial Intelligence

The projects related to our “Wet Line Detector” deployments have been a great experience for us. I can say a lot of warm words about how seamless & smooth cooperation with our client’s staff is. However, regarding the paper industry itself, it really is full of surprises. Prior to the first projects, I had never thought about using AI solutions there. After all, it is yet another great industry where AI has a lot to offer, be it predictive maintenance, Industry 4.0, Big Data analytic, Computer Vision, just to name a few. And we as a byteLAKE team are proud to add it to our AI offering for manufacturing, said Mariusz Kolanko, co-founder of byteLAKE

Humans and machines both make mistakes. Therefore, byteLAKE’s Cognitive Services have been designed to effectively bring out the best of both worlds.

byteLAKE continues collaboration with the industry leaders to develop further modules within the Cognitive Services offering as Artificial Intelligence has a huge potential for manufacturing automation and quickly delivers measurable benefits.

byteLAKE’s Wet Line Detector for Paper Industry and Cognitive Services (AI for Industry 4.0: Visual Inspection & Big Data Analytics) are available as a complete, end-to-end solution, ready for integration across industries.

byteLAKE’s AI solutions for Industry 4.0

The deployment is a straightforward process. We install our intelligent cameras that can monitor processes to detect anomalies, count products, read QR/bar codes, labels, inspect defects, etc. The cameras do not have to be mounted above the production line. Our algorithms have been trained to work with different perspectives and angles. Besides visual analytics, the solution can analyze data from various sources in order to find patterns, make predictions, or supports various decisions related to maintenance, conclusions based on historic data analytics, etc.

About byteLAKE

Artificial Intelligence for Industries. Products and Services.

• AI: enabling innovative, AI-powered automation and data-driven, proactive operations. Solutions for Fluid Dynamics, Manufacturing, Industry 4.0, Big Data, Document Processing, Restaurants, Construction Industry, Chemical Industry, Agriculture, Logistics, and more.

• HPC: accelerating time to results and adapting complex algorithms to many-CPU, GPU, FPGA architectures.

About byteLAKE’s Cognitive Services

byteLAKE’s Cognitive Services is a collection of AI models designed to address Industry 4.0 needs. Each AI model we designed and trained is razor-focused on specific industry tasks, therefore ensuring maximum accuracy. In essence, byteLAKE’s Cognitive Services focus primarily on the following two areas:

· AI-assisted visual inspection for efficient quality of products & process monitoring.

· AI-powered Big Data / IoT sensors data analytics to find trends, enable predictive maintenance, answer questions like why something happens, what will likely happen or to find the collective meaning of the data extracted from many sources.

There are a couple of special versions of Cognitive Services:

  • Wet Line Detector for Paper Mills: automates detection and measurements of the so-called wet line which occurs during paper production. This module helps reduce costs, expedite visual inspection and avoid unplanned downtime in paper mills. Available as a complete end-to-end solution, including edge components (hardware computing units, cameras), software license, integration services or guidelines, and post-deployment customer care. Read more:
  • AI for Restaurants: intelligent cameras that recognize food types and automate the billing process.

byteLAKE’s Cognitive Services are powered by byteLAKE’s brainello which is the AI-powered OCR to read text data i.e. invoices, QR/bar codes, labels/stickers etc. As brainello is AI-powered, it works without any templates or rules. In addition, it can learn and improve over time.

Complementary offering: byteLAKE’s CFD Suite, a collection of innovative AI Models for Computational Fluid Dynamics (CFD) acceleration. Learn more:


Cell: +48 505 322 282, +48 508 091 885


Cognitive Services team:

Read more in the related blog post series: Artificial Intelligence for Industry 4.0 (blog post series) — table of contents.



Marcin Rojek

Co Founder@byteLAKE | Turning Data Into Information for Manufacturing, Automotive, Paper,Chemical,Energy sectors | AI-accelerated CFD | Self-Checkout for Retail