Hold up any of the asamyuta hasta mudras to the camera, and the model I trained will tell you which it is, and give you a few uses!
Starting camera…
⚠️ The model is still being trained — check back soon!
Show a mudra to the camera to identify it.
0 samples
Hold the mudra clearly in the camera frame, then press Start Recording. Keep holding while it captures — aim for 30+ samples per mudra. The more samples, the better! When you're done, click Export JSON — it downloads mudra-training-data.json, ready to hand off and commit to public/mudra-training-data.json so the live site uses it too.