Related Experiment Video
Updated: Aug 16, 2026

Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
Image and attribute dataset for 16 464 post-consumer garments discarded or donated in Norway
Johan Berg Pettersen1, Solveig Aarak1, Edyy Vanessa Peña Benítez1
1Industrial Ecology Programme, Department of Energy and Process Engineering. Norwegian University of Science and Technology, Norway.
Abstract:
There is an increasing focus on the sustainability of clothes and textile consumption. European policymakers seek to increase resale and recovery of textiles, as recycled fiber material represents only a very minor proportion of total textiles production. Data for product and material composition and other qualities is in demand. The database presents image and attribute data for post-consumer clothes. Sample size is 16 464 garment items, equivalent to 4 157 kg of post-consumer clothes collected in 6 urban regions of Norway. Samples are sorted from households to "clothes, shoes and other textile products sorted for reuse" (Donation), and "dry and clean clothes and textile not for reuse" (Textile waste), collected at central recycling stations and distributed donation boxes. Samples undergo visual inspection of garment and product care label, and individual garment photography, to produce (1) a data table of garment attributes, and (2) a dataset of images. Sample attributes include product category; weight; a five-level score for degree-of-wear; brand; intended user; manufacturing method; number of layers; several other attributes that are numerical (velcro, metal and plastic buttons, metal and plastic zippers, metal rivets/eyelets/aglets, pearls, reflectors, and holes front and total) or Boolean (sequins, print, print >100 cm2, all-over print, other product parts, care label presence and readability, home-made, price tag, waterproof membrane). Care label information is used to record year and country of production, and fiber blend for individual layers of the garment. The data has multiple potential uses, among them the estimation of fiber blends, fiber recovery potential, resale potential and sorting for resale routines, and product lifetime estimation. Generally, the information can inspire and inform innovation in circular economy for textiles, inform extended producer responsibility (EPR) actors and direct sustainable textile policy.
Related Concept Videos
PPE Use in Healthcare Settings II: Doffing
PPE Use in Healthcare Settings I: Donning
Goodness-of-Fit Test
Data Collection II
Data Collection by Observations
An astronomer viewing the motion and brightness of stars in the sky and recording the data is an example of observational data collection. A botanist recording...
How Data are Classified: Categorical Data
Data are classified based on whether they are measurable or not. Categorical data cannot be measured; instead, it can be divided into categories. For example, if Y denotes a person's party affiliation, some examples of Y include...
