The vision tasks — detection, segmentation, tracking, recognition, OCR — and the models that power them.