You can deploy this model and start earning money today!

UWyo/
wildlife-north-american-wildlife

Name: wildlife-north-american-wildlife
Availability: InStock

north-american-wildlife

multi-species

cc-by-4.0

Views

No views yet

Model Card — North American Wildlife (26-class)

Single-stage object detection model covering 26 North American wildlife species, fine-tuned from the Ultralytics YOLO26s backbone (pretrained on COCO).

Model file: yolo26s_finetuned_26-wildlife-class_by_J.Gong_uwyo_2026-05-28.pt

Classes

ID	Common name	Latin name
0	Golden Eagle	Aquila chrysaetos
1	Pronghorn	Antilocapra americana
2	Bighorn Sheep	Ovis canadensis
3	American Bison	Bison bison
4	Mule Deer	Odocoileus hemionus
5	Elk / Wapiti	Cervus canadensis
6	Coyote	Canis latrans
7	Grizzly Bear	Ursus arctos horribilis
8	Gray Wolf	Canis lupus
9	Moose	Alces alces
10	American Pika	Ochotona princeps
11	Swift Fox	Vulpes velox
12	Mountain Lion	Puma concolor
13	North American River Otter	Lontra canadensis
14	American Black Bear	Ursus americanus
15	Bald Eagle	Haliaeetus leucocephalus
16	Red-tailed Hawk	Buteo jamaicensis
17	Osprey	Pandion haliaetus
18	Greater Sage-Grouse	Centrocercus urophasianus
19	Trumpeter Swan	Cygnus buccinator
20	North American Beaver	Castor canadensis
21	Common Raven	Corvus corax
22	Black-tailed Prairie Dog	Cynomys ludovicianus
23	American Badger	Taxidea taxus
24	Bobcat	Lynx rufus
25	Black-tailed Jackrabbit	Lepus californicus

Training Details

Property	Value
Base model	yolo26s.pt (COCO pretrained, Ultralytics)
Architecture	YOLO26s
Input size	640 × 640
Epochs	100
Optimizer	MuSGD, lr=0.002, momentum=0.9
Augmentation	mosaic=1.0, degrees=10°, scale=0.5, fliplr=0.5, hsv_h/s/v
Device	NVIDIA RTX 5000 Ada Generation (32 GB, CUDA 12.8)
Training date	2026-05-28
Author	Jian Gong, University of Wyoming

Dataset

Images sourced from iNaturalist (research-grade observations). Bounding boxes generated by MegaDetector v5a (confidence ≥ 0.15), then converted to YOLO format. Split 80 / 10 / 10 train / val / test.

Split	Images
train	5,917
val	727
test	762

Performance

Evaluated on the held-out validation set (best checkpoint).

Metric	Value
mAP50	0.9821
mAP50-95	0.9006

Per-class breakdown requires re-running training/04_evaluate.py --weights <this model>.

Usage

python

from ultralytics import YOLO
model = YOLO("models/north_american_wildlife/yolo26s_finetuned_26-wildlife-class_by_J.Gong_uwyo_2026-05-28.pt")
results = model.predict("image.jpg", conf=0.25)
for r in results:
    for box in r.boxes:
        print(model.names[int(box.cls)], float(box.conf))

Notes

Use this model as a general-purpose wildlife detector, or as a base for per-species fine-tuning (models/north_american_wildlife/ → species folder).
For deployment on Jetson Orin Nano, export to TensorRT FP16: yolo export model=<this file> format=engine half=True imgsz=640

UWyo/wildlife-north-american-wildlife