Issues

• Core AI semantic segmentation spends 3.8 ms per frame on a Swift argmax over logits

  bugenhancementtriage
  ultralytics/yolo-ios-app#322 · Filter by author glenn-jocher opened on Sep 18, 2026
  2

• Core AI preprocessing costs about 1.2 ms of CPU per frame and makes depth slower than Core ML

  ultralytics/yolo-ios-app#321 · Filter by author glenn-jocher opened on Sep 18, 2026

• Apple Core AI (.aimodel) support: opt-in now, criteria for making it the default

  embeddedenhancementtriage
  ultralytics/yolo-ios-app#320 · Filter by author glenn-jocher opened on Sep 18, 2026
  1

• Benchmark iOS inference-latency ideas

  enhancementtriage
  ultralytics/yolo-ios-app#313 · Filter by author glenn-jocher opened on Aug 17, 2026
  1