What GarmentScan Does
Send GarmentScan a photo of any clothing item — or its care and composition label — and it returns the full environmental profile of that garment: how much CO2 was emitted to produce it, how many litres of water were consumed from fibre to finished product, what its bill of materials looks like, how long it takes to produce and what it costs to manufacture, and a preview of its EU Digital Product Passport.
The tool is designed for fashion industry professionals, sustainability teams, researchers and conscious consumers who need actual numbers rather than vague sustainability claims.
How It Works
Image Classification and Two-Step Flow
Every image is first classified as a garment photo, a label photo, or neither. The bot then routes accordingly.
The recommended workflow is two-step: send a garment photo first, then optionally add a photo of the care and composition label. The garment photo identifies the type, construction complexity and estimated materials from visual cues. The label photo provides exact fibre percentages, country of origin, brand and care instructions. Together, they produce maximum accuracy. Either alone is sufficient for a useful result.
Garment Identification
The garment vision model (Qwen2.5-VL-72B, with multiple fallback models) analyses the clothing photo and extracts:
- Specific garment type (crew neck t-shirt, slim fit jeans, puffer jacket)
- Estimated material composition inferred from visual cues — sheen indicates polyester or nylon, matte indicates cotton, fuzzy texture indicates wool, stretch indicates elastane
- Fabric construction type (jersey knit, woven, denim, fleece) and estimated weight in GSM
- Trims: buttons, zippers, elastic, labels, rivets
- Colour and finish treatment: garment-dyed, printed, washed, embroidered
- Construction complexity: basic, moderate or complex
- Estimated weight in grams
Label Reading
When a label photo is provided, a separate vision pass reads the exact text: fibre percentages with full precision, country of origin, brand name, size, and care instructions including wash temperature, bleach, tumble dry and iron settings. Label data overrides estimates wherever available — exact composition is always more reliable than visual inference.
If the user knows the actual weight of the garment, they can type it in grams at any time and the entire calculation is rerun with the corrected figure.
The Environmental Calculations
CO2 Footprint
Carbon emissions are calculated stage by stage through the garment lifecycle:
- Fibre production: Emissions per kilogram vary significantly by material — cotton, polyester, wool, nylon and blends each carry different carbon intensities based on published lifecycle assessment data.
- Fabric production: Spinning, weaving or knitting, and finishing processes add emissions proportional to fabric weight and construction complexity.
- Dyeing and treatment: Chemical processes including dyeing, printing and special finishes carry their own CO2 and chemical load.
- Cut, make and trim (CMT): Factory production energy use, calculated from construction complexity and production time estimates.
- Transport: Based on country of origin and typical shipping routes to market.
The total is expressed in kg of CO2 equivalent, with equivalents provided for context: kilometres driven by car, and number of smartphone charges. A Show Math button reveals the exact calculation for every figure, with source citations for each emissions factor.
Water Usage
Water consumption is broken down by fibre type, dyeing process and finishing. Cotton is particularly water-intensive at the growing stage; synthetic fibres consume more water in chemical processing. The total is expressed in litres, with an equivalent number of showers for immediate comprehension.
Bill of Materials
The BOM lists every component of the garment — face fabric, lining, interlining, buttons, zippers, labels, thread, elastic, rivets — with quantities and individual weights. Fabric area in square metres and estimated cutting waste percentage are included, which are critical inputs for accurate material cost and impact calculations in a manufacturing context.
Production Data
Based on garment type and construction complexity, GarmentScan estimates:
- Total production time in minutes, broken down by operation
- CMT (Cut, Make, Trim) cost in USD at standard factory rates
- Machines required: sewing machines, overlockers, presses, specialist equipment
- Typical production country for this garment category
- Minimum order quantity for standard factory production
This data is useful for fashion brands costing new styles, for sustainability auditors benchmarking production efficiency, and for researchers studying the economics of garment manufacturing.
Ecobalyse Integration
GarmentScan queries the Ecobalyse API, the French government’s open-source tool for textile environmental impact assessment using the EU Product Environmental Footprint (PEF) methodology. Where Ecobalyse data is available for the garment’s profile, it is displayed alongside the in-house calculation — providing a second reference point using the official EU standard methodology. The climate change score in kg CO2e and the PEF score in micropoints are both reported.
EU Digital Product Passport Preview
The EU’s Digital Product Passport (DPP) regulation is coming into force for textiles as part of the Ecodesign for Sustainable Products Regulation (ESPR). It will require brands to provide structured environmental data for each product. GarmentScan generates a DPP preview with the fields that the regulation requires:
- Recyclability: Rated GREEN, YELLOW or RED with a specific reason and recommendation based on material composition and construction
- Disassembly: Which components are removable and difficulty rating
- Durability: Estimated wash cycles based on fabric type and fibre quality, with separate fabric and fibre durability ratings
- Substances of concern: Assessment of chemical risk from dyes, finishes and synthetic fibres
- DPP completeness score: How complete the passport would be with the available data, with a list of missing fields
For fashion brands preparing for DPP compliance, this preview identifies what data gaps need to be filled and what the regulatory picture looks like for existing products.
Greta Mode
The Greta Mode summary gives an immediate, colour-coded environmental verdict across three dimensions:
- Carbon: GREEN (low emissions), YELLOW (moderate), RED (high) with a brief explanation
- Water: GREEN, YELLOW or RED with context on which materials drive the usage
- Overall: A combined sustainability rating with a plain-language comment
This is the consumer-facing layer of the tool — an instant answer to the question of how environmentally costly a specific garment is, without requiring the user to interpret lifecycle assessment numbers.
Who GarmentScan Is For
Fashion Brands and Sustainability Teams
Rapid environmental assessment of new styles during the design phase — before sampling and production costs are committed. The DPP preview identifies compliance gaps early. The BOM and production data support costing and sustainability reporting in parallel.
Buyers and Sourcing Teams
Environmental profiling of garments during the sourcing process, with country-of-origin data and production cost estimates that complement commercial pricing.
Sustainability Auditors and Consultants
Structured, calculable environmental data for garments across a brand’s range. The Show Math transparency layer means every figure can be audited and traced to its source assumption.
Researchers and Academics
Rapid lifecycle assessment estimates for large garment datasets, with consistent methodology and transparent calculation. The Ecobalyse integration provides comparison against the EU PEF standard.
Conscious Consumers
A practical tool for understanding the environmental cost of individual purchasing decisions — not marketing copy, but actual calculated numbers from the garment in hand.
The Architecture
GarmentScan is a Telegram bot built on the same decision-support pipeline architecture as CrateVision and AutoValuta. Vision models handle image classification and information extraction. A rule-based calculator applies published lifecycle assessment factors to produce environmental metrics. The Ecobalyse API provides EU PEF reference data. A 72B LLM generates the Greta Mode commentary. All results are logged to a personal dashboard.
If you are building sustainability tooling for a fashion brand, a supply chain platform, or a consumer-facing application — or if you need to prepare for EU Digital Product Passport compliance — get in touch. We build custom AI pipelines for data extraction, environmental assessment and regulatory reporting.