Model Zoo¶

PyTorch-Wildlife provides a growing library of detection, classification, and bioacoustic models. All models load with a single line and download weights automatically.

Detection Models¶

MegaDetector V6¶

The latest generation of MegaDetector, trained on diverse global camera-trap datasets. Multiple architecture variants are available to trade off accuracy vs. speed vs. licensing.

Version	Architecture	License	Load with
`MDV6-yolov10-c`	YOLOv10 Compact	AGPL	`MegaDetectorV6(version="MDV6-yolov10-c")`
`MDV6-yolov10-e`	YOLOv10 Extra	AGPL	`MegaDetectorV6(version="MDV6-yolov10-e")`
`MDV6-yolov9-c`	YOLOv9 Compact	AGPL	`MegaDetectorV6(version="MDV6-yolov9-c")`
`MDV6-yolov9-e`	YOLOv9 Extra	AGPL	`MegaDetectorV6(version="MDV6-yolov9-e")`
`MDV6-mit-yolov9-c`	YOLOv9 Compact	MIT	`MegaDetectorV6MIT(version="MDV6-mit-yolov9-c")`
`MDV6-mit-yolov9-e`	YOLOv9 Extra	MIT	`MegaDetectorV6MIT(version="MDV6-mit-yolov9-e")`
`MDV6-apa-rtdetr-c`	RT-DETR Compact	Apache 2.0	`MegaDetectorV6Apache(version="MDV6-apa-rtdetr-c")`
`MDV6-apa-rtdetr-e`	RT-DETR Extra	Apache 2.0	`MegaDetectorV6Apache(version="MDV6-apa-rtdetr-e")`

from PytorchWildlife.models import detection as pw_detection

# Default (AGPL, YOLOv10)
detector = pw_detection.MegaDetectorV6()

# MIT-licensed YOLO
detector = pw_detection.MegaDetectorV6MIT(version="MDV6-mit-yolov9-e")

# Apache RT-DETR
detector = pw_detection.MegaDetectorV6Apache(version="MDV6-apa-rtdetr-e")

MegaDetector V5¶

The previous generation, widely deployed across conservation organizations. Uses YOLOv5.

detector = pw_detection.MegaDetectorV5()

For V5 model weights and earlier versions, see the archive branch of the Biodiversity repository.

Deepfaune Detector¶

Trained for European ecosystems. The first third-party camera-trap detection model integrated into PyTorch-Wildlife.

detector = pw_detection.DeepfauneDetector()

See the Deepfaune website for more details.

HerdNet¶

Point-based localization model for overhead and aerial imagery.

detector = pw_detection.HerdNet()

Classification Models¶

All classifiers can be paired with any detection model to build a detection + classification pipeline.

Model	Class	Geography	Species
AI4G Amazon Rainforest	`AI4GAmazonRainforest`	Amazon	~36 species
AI4G Snapshot Serengeti	`AI4GSnapshotSerengeti`	African savanna	~48 species
AI4G Opossum	`AI4GOpossum`	Americas	Opossum vs. non-opossum
Deepfaune	`DeepfauneClassifier`	Europe	~44 species
DFNE	`DFNE`	Northeastern North America	Fine-tuned Deepfaune

from PytorchWildlife.models import classification as pw_classification

classifier = pw_classification.AI4GAmazonRainforest()
classifier = pw_classification.AI4GSnapshotSerengeti()
classifier = pw_classification.DeepfauneClassifier()
classifier = pw_classification.DFNE()

Detection + Classification Pipeline¶

from PytorchWildlife.models import detection as pw_detection
from PytorchWildlife.models import classification as pw_classification

detection_model = pw_detection.MegaDetectorV6()
classification_model = pw_classification.AI4GAmazonRainforest()

# Detect, then classify crops
detection_result = detection_model.single_image_detection("image.jpg")
classification_result = classification_model.single_image_classification("image.jpg")

For a full pipeline demo, see the demo/detection_classification_pipeline_demo.py script.

Bioacoustic Models¶

from PytorchWildlife.models import bioacoustics as pw_bioacoustics

model = pw_bioacoustics.BioacousticsResnetClassifier()

For the full bioacoustic model zoo, see microsoft/MegaDetector-Acoustic.