Textile waste is one of the fastest-growing material streams in the global economy, and progress towards a circular textile economy depends on automating the sorting, separation and disassembly. Implementation of Artificial Intelligence (AI) has been proposed as a promising solution, but the literature on AI in textile recycling has not previously been organised around the operational decisions that an industrial recycling line must make. In this systematic review, 173 primary published studies between 1995 and 2026 were reviewed through a six-stage pipeline framework covering the detection of 1) textile type, 2) colour and pattern, 3) surface contamination, 4) functional reusability, 5) attachment removal and separability, and 6) material composition. Stage 4 was found to be saturated by manufacturing-oriented structural defect detection, which accounts for most of the corpus, while assessment of genuinely post-consumer garment condition remains almost as sparse as stage 5. Stages 1, 2, 3 and 6 are unevenly developed, and stage 5 is almost absent, with only one paper identified in the corpus reviewed here. Most technologies remain at the laboratory prototype stage, with limited industrial adoption. The highest-priority research direction is closure of the stage 5 gap, supported by post-consumer benchmarks and cost-effective near-infrared sensing.
No takes yet. Share an insight, caveat, or question.
Heidari et al. (2026) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: