Maria José Ferreira and Green Shoes 4.0 Consortium

In an era where digital transformation and collaboration is essential for business success, the Greenshoes 4.0 consortium contributes to driving technological evolution in the footwear sector by developing easy to use digital tools. With the active collaboration of industry leaders and research institutions, this Portugal 2020 R&D collaborative project is promoted by 15 companies covering the whole footwear value chain and eight R&D bodies with complementary capabilities.

It is shaping the future of fashion through the application of ecodesign, new materials and products, life cycle evaluation as well as statistics, machine learning techniques and advanced computer vision.  As explained by Maria José Ferreira, Director at CTCP (Centro Tecnológico do Calçado de Portugal) and Green Shoes 4.0 coordinator, the consortium has been working to drive the development of innovative services and applications to optimise digital activities and business models in footwear production, supply chain, retail and online sales.

Footwear traceability 

A simple footwear traceability tool has also been developed. This should be understood as being a first step to integrate traceability in the mindset and communication practices of producers and customers. The tool developed by CTCP with the support of the footwear and leathergoods companies participating in the consortium helps companies provide customers with information such as the type and origin of materials, percentage of recycled content, materials and product manufacturing locations and certifications.

The tool data system allows easy updating of this information by the company that is responsible for how the data will be collected, its accuracy and authenticity, and linking to each shoe's unique identifier. An appellative user-friendly interface allows customers to access traceability information easily through QR (quick response) codes or a search function on a website. Further research and complete and robust tools and methodologies are required to ensure validation of data and transparency in footwear traceability.

Forecasting sales

Forecasting sales of products is important for effective financial management, supply chain optimisation, production planning, marketing strategies, new product introductions, risk management and minimising waste from unsold products. Accurate sales forecasts enable businesses to make informed decisions, optimise resources and respond proactively to market dynamics, leading to improved efficiency, profitability and competitiveness.

In this regard, a demand forecasting system has been developed to respond to two distinct needs of the footwear sector. On the one hand, to forecast sales of products already on the market, using existing sales history to train the forecasting algorithm and anticipate market behaviour. On the other hand, to forecast sales of new products that are going to be introduced to the market for the first time without any previous history. In this case, the forecast is based on identifying a set of similar items with a known sales history and exploring their behaviour through trend and seasonality analysis.

The demand forecasting system supports production and planning by providing strategies to estimate, with a certain degree of accuracy, the demand values for a given period in the future. The system was developed by INESTEC, ISEP, AMF, OFICINAWARE and CTCP based on a set of statistical techniques and machine learning algorithms that allow it to effectively model trends, seasonality and events with their impact on time series data. The system has been tested and validated at an industrial level to help regulate the quantity and scheduling of materials purchasing and reduce overproduction that could result in unsold products and the resulting environmental and economic impacts.

Injection system tool

The footwear industry extensively uses the injection moulding process for attaching polyurethane (PU) soles directly to the shoe upper. This method offers several advantages, namely enhanced bonding between the sole and upper, and improved water resistance. Correct planning is important for maximising production efficiency while also maintaining quality, increasing output and meeting customer expectations. It also helps to optimise workflow, utilisation of equipment and moulds, material consumption, allocation of human resources, lead times and scalability, thus contributing to faster and more productive operations.

Injection moulding is also one of the most critical processes in footwear production due to the complexity associated with the injection machine itself and its operational constraints. These include availability of moulds by sizes, possible colour changes, and stabilisation times after injection, among others. In this sense, the production planning tool has been developed by the aforementioned organisations with two levels of decision making in mind. Firstly, a more strategic-tactical level of lot-sizing and workload distribution over time and, secondly, a more operational level of production sequencing by machine position.

The first level consists of determining a macro production plan with a time horizon of 6-10 weeks with the main objective of distributing the workload for each period (week), identifying the products and quantities to be produced, stock levels and resources required to minimise production and stock costs and delays. Specifically, this strategic planning tool has been developed using integer programming mathematical models, with the proposition that only 80% of machine capacity should be used from the second week onwards to ensure flexibility to integrate new orders into the plan.

The second level consists of defining a weekly production plan, determining the allocation of production orders and their quantities to the various positions of the injection machine, bearing in mind the associated restrictions (possible colour changes, mould changes, etc.). In this instance, the main objective is to maximise machine occupancy rate and minimise processing time. The sequencing tool is based on a constructive heuristic, integrating randomisation components to ensure a more efficient exploration of the various possible solutions.

Visual search tool

One of the solutions implemented was a visual search service designed to allow customers to find footwear products based on images. Neural networks specialised in classifying colour, gender, shoe type and heel size were configured and trained for this purpose. This function provides a convenient and efficient way to find specific products according to the user's preferences. In this project, CCG, KYAIA, KSI, Overcube and CTCP have also sought to mitigate the need for a human operator to use technical image editing tools to remove the backgrounds in photos of shoes that are produced for a catalogue, as well as the time associated with this manual process. 

To this end, a shoe segmentation service has been developed for semi-automatic catalogue creation. On the one hand, a segmentation neural network called U2-Net has been implemented, which builds a saliency map of the image, highlighting in a rigorous way the object in the foreground (e.g. shoe), which favours the possibility of achieving a smooth and precise separation between the footwear and the background (or second-plane). In addition, the previous service (Visual Search Tool) is used to extract the characteristics of the shoe to be catalogued (gender, colour, heel and type) and to fill in a shoe registration form with proposals automatically that the user can modify if they wish. With this tool, the process of creating catalogues has been simplified and significantly streamlined. 

ALL CREDITS CTCP

Share this story: