This repository represents Ultralytics open-source research into future object detection methods, and incorporates lessons learned and best practices evolved over thousands of hours of training and evolution on anonymized client datasets. All code and models are under active development, and are subject to modification or deletion without notice. Use at your own risk.

YOLOv5-P5 640 Figure (click to expand)

Figure Notes (click to expand)
  • 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.
  • Reproduce by python test.py --task study --data coco.yaml --iou 0.7 --weights yolov3.pt yolov3-spp.pt yolov3-tiny.pt yolov5l.pt

Branch Notice

The ultralytics/yolov3 repository is now divided into two branches:

$ git clone https://github.com/ultralytics/yolov3  # master branch (default)
$ git clone https://github.com/ultralytics/yolov3 -b archive  # archive branch

Pretrained Checkpoints

Model size
V100 (ms)
640 (B)
YOLOv3-tiny 640 17.6 17.6 34.8 1.2 8.8 13.2
YOLOv3 640 43.3 43.3 63.0 4.1 61.9 156.3
YOLOv3-SPP 640 44.3 44.3 64.6 4.1 63.0 157.1
YOLOv5l 640 48.2 48.2 66.9 3.7 47.0 115.4
Table Notes (click to expand)
  • APtest denotes COCO test-dev2017 server results, all other AP results denote val2017 accuracy.
  • AP values are for single-model single-scale unless otherwise noted. Reproduce mAP by python test.py --data coco.yaml --img 640 --conf 0.001 --iou 0.65
  • SpeedGPU averaged over 5000 COCO val2017 images using a GCP n1-standard-16 V100 instance, and includes FP16 inference, postprocessing and NMS. 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).


Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt



YOLOv3 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):


detect.py runs inference on a variety of sources, downloading models automatically from the latest YOLOv3 release and saving results to runs/detect.

$ python detect.py --source 0  # webcam
                            file.jpg  # image 
                            file.mp4  # video
                            path/  # directory
                            path/*.jpg  # glob
                            'https://youtu.be/NUsoVlDFqZg'  # YouTube video
                            'rtsp://example.com/media.mp4'  # RTSP, RTMP, HTTP stream

To run inference on example images in data/images:

$ python detect.py --source data/images --weights yolov3.pt --conf 0.25

PyTorch Hub

To run batched inference with YOLOv3 and PyTorch Hub:

import torch

# Model
model = torch.hub.load('ultralytics/yolov3', 'yolov3')  # or 'yolov3_spp', 'yolov3_tiny'

# Image
img = 'https://ultralytics.com/images/zidane.jpg'

# Inference
results = model(img)
results.print()  # or .show(), .save()


Run commands below to reproduce results on COCO dataset (dataset auto-downloads on first use). Training times for YOLOv3/YOLOv3-SPP/YOLOv3-tiny are 6/6/2 days on a single V100 (multi-GPU times faster). Use the largest --batch-size your GPU allows (batch sizes shown for 16 GB devices).

$ python train.py --data coco.yaml --cfg yolov3.yaml      --weights '' --batch-size 24
                                         yolov3-spp.yaml                            24
                                         yolov3-tiny.yaml                           64