yolov3-pytorch 0.1.1

YOLOv3 in PyTorch.

YOLOv3-PyTorch

Contents

• Introduction
• Getting Started
  • Requirements
  • Local Install
• Download weights
• Download datasets
• How Test and Train
  • Test yolov3_tiny_voc model
  • Train yolov3_tiny_voc model
  • Resume train yolov3_tiny_voc model
• Result
• Contributing
• Credit
  • YOLOv3: An Incremental Improvement

Introduction

This repository contains an op-for-op PyTorch reimplementation of YOLOv3: An Incremental Improvement.

Getting Started

Requirements

• Python 3.10+
• PyTorch 2.0.0+
• CUDA 11.8+
• Ubuntu 22.04+

From PyPI

pip3 install yolov3_pytorch -i https://pypi.org/simple

Local Install

git clone https://github.com/Lornatang/YOLOv3-PyTorch.git
cd YOLOv3-PyTorch
pip3 install -r requirements.txt
python3 setup.py install

Inference (TODO)

# Download YOLOv3-tiny model weights to `./results/pretrained_models`
wget .
python3 ./tools/detect.py

All pretrained model weights

• Google Driver
• Baidu Driver

How Test and Train

Both training and testing only need to modify the train_config.py or test_config.py file.

Test yolov3_tiny_voc model

Modify the test_config.py file.

• line 18: model_arch_name change to yolov3_tiny_voc.
• line 34: test_dataset_config_path change to ./data/voc.data.
• line 38: model_weights_path change to ./results/pretrained_models/YOLOv3_tiny-COCO.weights.

python3 test.py

Train yolov3_tiny_voc model

Modify the train_config.py file.

• line 18: model_arch_name change to yolov3_tiny_voc.
• line 58: upscale_factor change to ./data/voc.data.

python3 train.py

Resume train yolov3_tiny_voc model

Modify the train_config.py file.

• line 18: model_arch_name change to yolov3_tiny_voc.
• line 58: upscale_factor change to ./data/voc.data.
• line 74: resume_model_weights_path change to f"./samples/YOLOv3_tiny-VOC0712/epoch_xxx.pth.tar".

python3 train.py

Result

Source of original paper results: https://arxiv.org/pdf/1804.02767v1.pdf

In the following table, the mAP value in () indicates the result of the project, and - indicates no test.

Model	Train dataset	Test dataset	Size	mAP
yolov3_tiny_prn_voc	VOC07+12 trainval	VOC07 test	416	-(56.4)
yolov3_tiny_voc	VOC07+12 trainval	VOC07 test	416	-(58.8)
yolov3_voc	VOC07+12 trainval	VOC07 test	416	-(79.0)
yolov3_spp_voc	VOC07+12 trainval	VOC07 test	416	-(75.3)
mobilenetv1_voc	VOC07+12 trainval	VOC07 test	416	-(66.0)
mobilenetv2_voc	VOC07+12 trainval	VOC07 test	416	-(69.3)
mobilenetv3_small_voc	VOC07+12 trainval	VOC07 test	416	-(53.8)
mobilenetv3_large_voc	VOC07+12 trainval	VOC07 test	416	-(71.1)
alexnet_voc	VOC07+12 trainval	VOC07 test	416	-(56.4)
vgg16_voc	VOC07+12 trainval	VOC07 test	416	-(74.5)

# Download `YOLOv3_tiny-VOC0712-d24f2c25.pth.tar` weights to `./results/pretrained_models`
# More detail see `README.md<Download weights>`
python3 ./detect.py

Output1:

Output2:

Loaded `` pretrained model weights successfully.
image 1/2 data/examples/dog.jpg: 480x608 1 bicycle, 1 car, 1 dog, 
image 2/2 data/examples/person.jpg: 416x608 1 dog, 1 person, 1 sheep,

Contributing

If you find a bug, create a GitHub issue, or even better, submit a pull request. Similarly, if you have questions, simply post them as GitHub issues.

I look forward to seeing what the community does with these models!

Credit

YOLOv3: An Incremental Improvement

Joseph Redmon, Ali Farhadi

Abstract
We present some updates to YOLO! We made a bunch of little design changes to make it better. We also trained this new network that’s pretty swell. It’s a little bigger than last time but more accurate. It’s still fast though, don’t worry. At 320 × 320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3 is quite good. It achieves 57.9 AP50 in 51 ms on a Titan X, compared to 57.5 AP50 in 198 ms by RetinaNet, similar performance but 3.8× faster. As always, all the code is online at https://pjreddie.com/yolo/.

[Paper] [Project Webpage] [Authors' Implementation]

@article{yolov3,
  title={YOLOv3: An Incremental Improvement},
  author={Redmon, Joseph and Farhadi, Ali},
  journal = {arXiv},
  year={2018}
}