Precise event spotting (PES) from pixels

You want to…	Read
Train a first model this afternoon, with every choice made for you	Quickstart — three sessions, one camera, plain E2E-Spot on pixels only
Understand every stage: the three ways to spot an event, the config, durations, confidence, MSAGSM	Full guide
Know how a curve becomes an event: one per class per trial, when a class is absent, several per trial	From probabilities to events
Add the pose features you already have as a second input beside the pixels	Pixels + Pose (extra features)
Look at all configs	Config reference

Whichever route you take, the predictions are a labels TSV the GUI opens with File ▸ Import labels…, and curation is where you review them.

• Quickstart
• Full guide
• From probabilities to events
• Pixels + pose
• Config reference