10.5281/zenodo.4435632
https://zenodo.org/records/4435632
oai:zenodo.org:4435632
Glenn Jocher
Glenn Jocher
@ultralytics
Yonghye Kwon
Yonghye Kwon
MarkAny
guigarfr
guigarfr
perry0418
perry0418
State Grid Electric Power Research Institute
Josh Veitch-Michaelis
Josh Veitch-Michaelis
University of Wisconsin-Madison
Ttayu
Ttayu
Image Algorigthm Engineer
Daniel Suess
Daniel Suess
@silverpond
Fatih Baltacı
Fatih Baltacı
Gabriel Bianconi
Gabriel Bianconi
Scalar Research
IlyaOvodov
IlyaOvodov
Elvees NeoTek JSC elveesneotek.ru
Marc
Marc
e96031413
e96031413
NCTU College of AI
Chang Lee
Chang Lee
Dustin Kendall
Dustin Kendall
Continental
Falak
Falak
Infocusp
Francisco Reveriano
Francisco Reveriano
Duke Applied Machine Laboratory
FuLin
FuLin
GoogleWiki
GoogleWiki
Jason Nataprawira
Jason Nataprawira
Ritsumeikan University
Jeremy Hu
Jeremy Hu
LinCoce
LinCoce
LukeAI
LukeAI
NanoCode012
NanoCode012
NirZarrabi
NirZarrabi
Oulbacha Reda
Oulbacha Reda
Polytechnique Montréal
Piotr Skalski
Piotr Skalski
@VirtusLab
SergioSanchezMontesUAM
SergioSanchezMontesUAM
Shiwei Song
Shiwei Song
Thomas Havlik
Thomas Havlik
Timothy M. Shead
Timothy M. Shead
Sandia National Laboratories
ultralytics/yolov3: v9.1 - YOLOv5 Forward Compatibility Updates
Zenodo
2021
2021-01-13
https://github.com/ultralytics/yolov3/tree/v9.1
10.5281/zenodo.2624707
v9.1
Other (Open)
This release is a minor update implementing numerous bug fixes, feature additions and performance improvements from https://github.com/ultralytics/yolov5 to this repo. Models remain unchanged from v9.0 release.
Branch Notice
The ultralytics/yolov3 repository is now divided into two branches:
Master branch: Forward-compatible with all YOLOv5 models and methods (recommended).
$ git clone https://github.com/ultralytics/yolov3 # master branch (default)
Archive branch: Backwards-compatible with original darknet *.cfg models (⚠️ no longer maintained).
$ git clone -b archive https://github.com/ultralytics/yolov3 # archive branch
<img src="https://user-images.githubusercontent.com/26833433/100382066-c8bc5200-301a-11eb-907b-799a0301595e.png" width="800">
** GPU Speed measures end-to-end time per image averaged over 5000 COCO val2017 images using a V100 GPU with batch size 32, and includes image preprocessing, PyTorch FP16 inference, postprocessing and NMS. EfficientDet data from google/automl at batch size 8.
Pretrained Checkpoints
Model
AP<sup>val</sup>
AP<sup>test</sup>
AP<sub>50</sub>
Speed<sub>GPU</sub>
FPS<sub>GPU</sub>
params
FLOPS
YOLOv3
43.3
43.3
63.0
4.8ms
208
61.9M
156.4B
YOLOv3-SPP
44.3
44.3
64.6
4.9ms
204
63.0M
157.0B
YOLOv3-tiny
17.6
34.9
34.9
1.7ms
588
8.9M
13.3B
AP<sup>test</sup> denotes COCO test-dev2017 server results, all other AP results denote val2017 accuracy. All AP numbers are for single-model single-scale without ensemble or TTA. Reproduce mAP by python test.py --data coco.yaml --img 640 --conf 0.001 --iou 0.65
Speed<sub>GPU</sub> averaged over 5000 COCO val2017 images using a GCP n1-standard-16 V100 instance, and includes image preprocessing, FP16 inference, postprocessing and NMS. NMS is 1-2ms/img. Reproduce speed by python test.py --data coco.yaml --img 640 --conf 0.25 --iou 0.45 All checkpoints are trained to 300 epochs with default settings and hyperparameters (no autoaugmentation).
Test Time Augmentation (TTA) runs at 3 image sizes. Reproduce TTA** by python test.py --data coco.yaml --img 832 --iou 0.65 --augment
Requirements
Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:
$ pip install -r requirements.txt