This is a pip package compatible adaptation of the official YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information reposistory.
MS COCO
Model | Test Size | APval | AP50val | AP75val | Param. | FLOPs |
---|---|---|---|---|---|---|
YOLOv9-T | 640 | 38.3% | 53.1% | 41.3% | 2.0M | 7.7G |
YOLOv9-S | 640 | 46.8% | 63.4% | 50.7% | 7.1M | 26.4G |
YOLOv9-M | 640 | 51.4% | 68.1% | 56.1% | 20.0M | 76.3G |
YOLOv9-C | 640 | 53.0% | 70.2% | 57.8% | 25.3M | 102.1G |
YOLOv9-E | 640 | 55.6% | 72.8% | 60.6% | 57.3M | 189.0G |
Expand
Custom training: WongKinYiu/yolov9#30 (comment)
ONNX export: WongKinYiu/yolov9#2 (comment) WongKinYiu/yolov9#40 (comment) WongKinYiu/yolov9#130 (comment)
ONNX export for segmentation: WongKinYiu/yolov9#260 (comment)
TensorRT inference: WongKinYiu/yolov9#143 (comment) WongKinYiu/yolov9#34 (comment) WongKinYiu/yolov9#79 (comment) WongKinYiu/yolov9#143 (comment)
QAT TensorRT: WongKinYiu/yolov9#327 (comment) WongKinYiu/yolov9#253 (comment)
OpenVINO: WongKinYiu/yolov9#164 (comment)
C# ONNX inference: WongKinYiu/yolov9#95 (comment)
C# OpenVINO inference: WongKinYiu/yolov9#95 (comment)
OpenCV: WongKinYiu/yolov9#113 (comment)
Hugging Face demo: WongKinYiu/yolov9#45 (comment)
CoLab demo: WongKinYiu/yolov9#18
ONNXSlim export: WongKinYiu/yolov9#37
YOLOv9 ROS: WongKinYiu/yolov9#144 (comment)
YOLOv9 ROS TensorRT: WongKinYiu/yolov9#145 (comment)
YOLOv9 Julia: WongKinYiu/yolov9#141 (comment)
YOLOv9 MLX: WongKinYiu/yolov9#258 (comment)
YOLOv9 StrongSORT with OSNet: WongKinYiu/yolov9#299 (comment)
YOLOv9 ByteTrack: WongKinYiu/yolov9#78 (comment)
YOLOv9 DeepSORT: WongKinYiu/yolov9#98 (comment)
YOLOv9 counting: WongKinYiu/yolov9#84 (comment)
YOLOv9 face detection: WongKinYiu/yolov9#121 (comment)
YOLOv9 segmentation onnxruntime: WongKinYiu/yolov9#151 (comment)
Comet logging: WongKinYiu/yolov9#110
MLflow logging: WongKinYiu/yolov9#87
AnyLabeling tool: WongKinYiu/yolov9#48 (comment)
AX650N deploy: WongKinYiu/yolov9#96 (comment)
Conda environment: WongKinYiu/yolov9#93
AutoDL docker environment: WongKinYiu/yolov9#112 (comment)
To install yolov9 as a package from PyPI:
pip install yolov9py
The package comes with an additional load utility:
from yolov9.load import load_model
model_cfg_path = ...
ckpt_path = ...
n_classes = 3
n_channels = 3
# This returns a nn.Module torch model
model = load_model(model_cfg_path, weights=ckpt_path, classes=n_classes, channels=n_channels)
@article{wang2024yolov9,
title={{YOLOv9}: Learning What You Want to Learn Using Programmable Gradient Information},
author={Wang, Chien-Yao and Liao, Hong-Yuan Mark},
booktitle={arXiv preprint arXiv:2402.13616},
year={2024}
}
@article{chang2023yolor,
title={{YOLOR}-Based Multi-Task Learning},
author={Chang, Hung-Shuo and Wang, Chien-Yao and Wang, Richard Robert and Chou, Gene and Liao, Hong-Yuan Mark},
journal={arXiv preprint arXiv:2309.16921},
year={2023}
}