Smart Garment Sorting System for Recycling

In order to effectively reuse used clothes, the clothing needs to be sorted so that recyclers can understand the types and materials of the clothing and adopt appropriate recycling methods. Conventional practice relies on manual sorting, which is labour intensive and carries concerns about hygiene. Therefore, The Hong Kong Research Institute of Textiles and Apparel (HKRITA) has developed the Smart Garment Sorting System for Recycling which combines artificial intelligence and image analysis to accurately categorise used garments.

Flexible Modules for Business Customisation

The smart garment sorting system comprises four core modules that can recognise garment type, fabric construction, material and colour. The four modules can be operated as stand-alone units or arranged in different orders to carry out specific sorting tasks and business operations. It makes use of a conveyor path to connect all the modules and robotic arms to pick up garments, further enhancing automation and recognition accuracy. Each individual module is also optimised to speed up recognition. For instance, the garment type can be identified even when the garment is hung in a relaxed arbitrary manner, thus saving time to stretch the garment to expose particular features.

Using robotic arms to pick up garments can reduce manual operation

A garment even hanging in a relaxed arbitrary manner can be identified

 

Adopting AI and Image Analysis for Higher Sorting Accuracy

The system embeds a self-developed database, allowing the modules to carry out bulk garments sorting through AI-based analysis technology and generate a full set of results in eight seconds. Aided with image analysis technology, the fabric construction recognition module can identify different fabric structures such as woven, knitted or nonwoven with a classification of accuracy over 95%. The material and colour classification modules also achieve an accuracy of over 90% by using hyperspectral cameras. The four modules operate based on commands sent from the central system. If an uncommon combination of garment type and material is detected, the system can prompt the user to confirm the accuracy of the classification.

Industrial Scale Upgrade to Speed up Textile Recycling

HKRITA will set up this garment sorting system in Open Lab where it can work in conjunction with other recycling facilities to perform textile recycling at scale. This will speed up the entire recycling process, achieving a closed-loop fashion manufacturing.

The system showcased in the 48^th International Exhibition of Inventions Geneva in April

This technology garnered a Gold Medal in the 48^th International Exhibition of Inventions Geneva

Mr Raymond Chiu (right), Senior Research Manager of HKRITA, introduced the system operation and its features to the local media in the media briefing session in May

Project video