How does YOLO learn, then find an object?

How does YOLO learn, then find an object? A data-flow diagram generated by Archify. 01 / Observe 02 / Make answers 03 / Train 04 / Check + new input 05 / Use Photo set · Different views · 01 / Observe · training input Photo set Different views training input Webcam feed · Frames not seen before · 04 / Check + new input · use input Webcam feed Frames not seen before use input Answer boxes · Name + location · 02 / Make answers · labels Answer boxes Name + location labels Dataset split · train / validation / test · 02 / Make answers · fair check Dataset split train /validation / test fair check YOLO training · Reduce prediction error · 03 / Train · repeat YOLO training Reduce prediction error repeat Validation · Test on unseen photos · 04 / Check + new input · overfit check Validation Test on unseen photos overfit check Trained model · best.pt weights · 04 / Check + new input · saved Trained model best.pt weights saved Detections · Box + name + confidence · 05 / Use · output Detections Box + name + confidence output Tello / RoboMaster · Separate safety extension · 05 / Use · later Tello / RoboMaster Separate safetyextension later Human adds answers annotated images Shuffle and separate dataset split Training subset training data Check each epoch metrics Best weights checkpoint Inference model output New frames live pixels Candidate after safety checks future extension Legend primary data async batch data store data flow