YOLO11-IRIS EMME-v2 Results (v0.1)

Final results for YOLO11-IRIS (EMME-v2) — version v0.1 This release archives the evaluation results, figures, and metrics from the integrated YOLO11 and EMME eye-tracking model. The results correspond to the experiments reported in the manuscript “Advancing Eye Analysis and Eye-Tracking with Ultralytics YOLO11 and EMME.” All experiments were performed with fixed random seeds and reproducible configurations. • Expected metrics: mAP@0.5 ≈ 0.995, precision ≈ 0.890, recall ≈ 0.830 • Latency: 35–50 ms per frame (~20–28 FPS) • License: MIT • Repository (code): https://github.com/seungjin-data/Advancing-Eye-Analysis-and-Eye-Tracking-with-Ultralytics-YOLO11 The uploaded figures (BoxPR_curve.png, confusion_matrix.png, and confusion_matrix_normalized.png) visualize detection accuracy, class confusion, and performance trade-offs. Quick start (reproduction):```bashpip install -r requirements.txtpython scripts/eval.py --cfg configs/eye_yolo11.yaml --weights <this_zenodo_artifact_or_local_path> --seed 42