Screen → camera, end to end
One container that takes you from no model to bytes sent over a screen: build a dataset (synthetic frames + real phone photos), train a YOLO26 pose detector, verify it, then use it in the S2C app.
The pipeline
–This box
loading…
Do this next
Recent activity
full guide →loading…
1 · Dataset
Synthetic frames, real captures from the phone, and a merge step. Inspect labels before you spend CPU on training.
labelled images
–
2 · Train
Fine-tune yolo26n-pose as a background job,
with the log streaming into the page.
runs
–
3 · Use it
The S2C app: play a file out of the laptop screen, read it back with the phone camera.
active model
–