Downloads · 30 days
97
27% of all-time downloads
sailswarm-yolo-audited
NexusDwin/sailswarm-yolo-audited
sailswarm-yolo-audited is a object detection model from NexusDwin. Use it when you need objects located in an image. It is set up for ultralytics. The card lists the license as other.
Ultralytics YOLOv8/YOLO11 detectors fine-tuned on the human-audited fisheye frames of the SailSwarm obstacle-detection corpus (Lake Constance, undistorted 864×648 fisheye; classes boat, buoy, duck, other, person, stru…
All-time downloads
359
Public
Repo size
286 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt172 MB · 60%
At a glance
- Task
- Object Detection
- Library
- ultralytics
- License
- other
- Access
- Public
- Created
- Sep 9, 2026
- Updated
- Sep 15, 2026
- SHA
- 3b3326ce
Model card
From the Hugging Face model README
SailSwarm YOLO, on-domain fine-tunes on the audited Konstanz frames
Ultralytics YOLOv8/YOLO11 detectors fine-tuned on the human-audited fisheye
frames of the SailSwarm obstacle-detection corpus (Lake Constance, undistorted
864×648 fisheye; classes boat, buoy, duck, other, person, structure, see
data.yaml). They supersede NexusDwin/sailswarm-yolov8n-konstanz (June 2026,
LaRS-era labels).
| weights | val (129 images, 1779 boxes) P / R / mAP50 / mAP50-95 | note |
|---|---|---|
weights/yolov8n_audited_best.pt | 0.440 / 0.415 / 0.454 / 0.268 | trained on every audited frame |
weights/yolov8s_audited_best.pt | 0.477 / 0.460 / 0.462 / 0.295 | (+ seeds s1, s2; 11n/11s siblings) |
weights/yolov8n_audited_night_best.pt / yolov8s_audited_night_best.pt | 0.430 / 0.398 / 0.418 / 0.254 and 0.471 / 0.469 / 0.447 / 0.284 (174 val images) | added 2026-09-15: audited corpus + the 2026-09-08 session audits (430 frames, 249 at night), 100 epochs at 864; runs/*/results.csv. Trained after the ICRA submission's holdout was frozen, so NOT the detector behind any paper number |
weights/yolov8n_noholdout_best.pt | 0.704 / 0.383 / 0.428 / 0.254 (68 held-out images, 1009 boxes) | leak-free: the audited holdout recordings removed from training; this is the detector behind the typed-evidence channel of the fusion scorer (GBT + typed AP 0.927 vs 0.874 without, 8 of 8 recordings) |
onnx/yolov8n_audited_640.onnx / _864.onnx and the v8s pair are the exports the box
runs (deploy/shadow_mode.sh typed; YOLOv8n-640 = 1.5 fps on a Raspberry Pi 4, 2
threads). Night-person use: v8s as a confirm-gate rejects 96 % of motor-as-person
teacher false positives while keeping 82 % of persons (2026-09-02).
Training logs: logs_yolo_*.log, yolo_noholdout_train.log. Recipe:
scripts/gpu_finetune/train_yolo.py in the SailSwarm-ObstacleDetection repo.
- Task
- Object Detection
- Library
- ultralytics
- License
- other
- Created
- Sep 9, 2026
- Updated
- Sep 15, 2026