The document discusses embedding data in knitted textiles, simulating knitting digitally, reverse engineering knitting patterns from existing knits, and developing a machine learning model to recognize stitch patterns in photographs of knitting. It describes collecting a dataset of knitting images, manually labeling them with their stitch patterns, training a deep learning model on the images, and deploying the model as a web application for stitch pattern recognition. Key challenges discussed include obtaining a large enough training set and achieving high accuracy on the validation set after multiple rounds of training.
5. @web_goddess
Overview
• Embedding data in textiles
• Knitting machines and DSLs
• Simulating knitting and digital manipulation
• Reverse engineering and forensic knitting
• My KnittingML project
• Lessons from the knitting community
49. @web_goddess
The plan
1. Get training data
2. Label training data
3. Train ML model
4. Deploy ML model
5. Build web frontend for inference
6. Dance on the grave of 1500-year-old knitting industry 💃