Introduction

YOLO RTX is an anchor-free and oriented object detection version of YOLO, with a simpler design but better performance! It aims to bridge the gap between research and industrial communities. For more YOLOX details, please refer to our report on Arxiv.

This repo is an implementation of PyTorch version YOLOX and add CSL rotated branch.

Quick Start

Installation

Step1. Install YOLOX.

git clone [email protected]:Megvii-BaseDetection/YOLOX.git
cd YOLOX
pip3 install -U pip && pip3 install -r requirements.txt
pip3 install -v -e .  # or  python3 setup.py develop

Step2. Install pycocotools.

pip3 install cython; pip3 install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'

Demo

Step1. Download a pretrained model from the benchmark table.

Step2. Use either -n or -f to specify your detector's config. For example:

python tools/demo.py image -n yolox-s -c /path/to/your/yolox_s.pth --path assets/dog.jpg --conf 0.25 --nms 0.45 --tsize 640 --save_result --device [cpu/gpu]

or

python tools/demo.py image -f exps/default/yolox_s.py -c /path/to/your/yolox_s.pth --path assets/dog.jpg --conf 0.25 --nms 0.45 --tsize 640 --save_result --device [cpu/gpu]

Demo for video:

python tools/demo.py video -n yolox-s -c /path/to/your/yolox_s.pth --path /path/to/your/video --conf 0.25 --nms 0.45 --tsize 640 --save_result --device [cpu/gpu]

Reproduce our results on COCO

Step1. Prepare COCO dataset

cd <YOLOX_HOME>
ln -s /path/to/your/COCO ./datasets/COCO

Step2. Reproduce our results on COCO by specifying -n:

python tools/train.py -n yolox-s -d 8 -b 64 --fp16 -o [--cache]
                         yolox-m
                         yolox-l
                         yolox-x

• -d: number of gpu devices
• -b: total batch size, the recommended number for -b is num-gpu * 8
• --fp16: mixed precision training
• --cache: caching imgs into RAM to accelarate training, which need large system RAM.

When using -f, the above commands are equivalent to:

python tools/train.py -f exps/default/yolox_s.py -d 8 -b 64 --fp16 -o [--cache]
                         exps/default/yolox_m.py
                         exps/default/yolox_l.py
                         exps/default/yolox_x.py

Multi Machine Training

We also support multi-nodes training. Just add the following args:

• --num_machines: num of your total training nodes
• --machine_rank: specify the rank of each node

Suppose you want to train YOLOX on 2 machines, and your master machines's IP is 123.123.123.123, use port 12312 and TCP.
On master machine, run

python tools/train.py -n yolox-s -b 128 --dist-url tcp://123.123.123.123:12312 --num-machines 2 --machine-rank 0

On the second machine, run

python tools/train.py -n yolox-s -b 128 --dist-url tcp://123.123.123.123:12312 --num-machines 2 --machine-rank 1

Evaluation

We support batch testing for fast evaluation:

python tools/eval.py -n  yolox-s -c yolox_s.pth -b 64 -d 8 --conf 0.001 [--fp16] [--fuse]
                         yolox-m
                         yolox-l
                         yolox-x

• --fuse: fuse conv and bn
• -d: number of GPUs used for evaluation. DEFAULT: All GPUs available will be used.
• -b: total batch size across on all GPUs

To reproduce speed test, we use the following command:

python tools/eval.py -n  yolox-s -c yolox_s.pth -b 1 -d 1 --conf 0.001 --fp16 --fuse
                         yolox-m
                         yolox-l
                         yolox-x

Tutorials

• Training on custom data
• Manipulating training image size

Deployment

1. MegEngine in C++ and Python
2. ONNX export and an ONNXRuntime
3. TensorRT in C++ and Python
4. ncnn in C++ and Java
5. OpenVINO in C++ and Python

Third-party resources

• The ncnn android app with video support: ncnn-android-yolox from FeiGeChuanShu
• YOLOX with Tengine support: Tengine from BUG1989
• YOLOX + ROS2 Foxy: YOLOX-ROS from Ar-Ray
• YOLOX Deploy DeepStream: YOLOX-deepstream from nanmi
• YOLOX ONNXRuntime C++ Demo: lite.ai from DefTruth
• Converting darknet or yolov5 datasets to COCO format for YOLOX: YOLO2COCO from Daniel

Cite YOLOX

If you use YOLOX in your research, please cite our work by using the following BibTeX entry:

 @article{yolox2021,
  title={YOLOX: Exceeding YOLO Series in 2021},
  author={Ge, Zheng and Liu, Songtao and Wang, Feng and Li, Zeming and Sun, Jian},
  journal={arXiv preprint arXiv:2107.08430},
  year={2021}
}

Install

sudo apt-get install swig -y

conda create --name yolox python=3.7 -y

conda activate yolox

pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.org/whl/torch_stable.html

pip install -U pip && pip install -r requirements.txt

pip install -v -e .  # or  python3 setup.py develop

pip install cython; pip install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'

cd DOTA_devkit

swig -c++ -python polyiou.i

python setup.py build_ext --inplace

cd ../yolox/utils/iou

swig -c++ -python polyiou.i

python setup.py build_ext --inplace

cd ../Rotated_IoU/cuda_op

python setup.py install

export TORCH_CUDA_ARCH_LIST="7.5"

DOTA datasets

PATH

cd DOTA_devkit

Step 1. Split

python ImgSplit_multi_process.py

Step 2. DOTA format to YOLO format

python YOLO_Transform.py

Step 3. YOLO format to COCO format

python darknet2coco.py  --data_path YOLO2COCO/gen_config.data

train

python tools/train.py -f exps/rotation/yolortx_m.py -d 1 -b 4 --fp16

demo

python tools/detect.py -f exps/rotation/yolortx_m.py -c YOLOX_outputs/yolortx_m/epoch_300_ckpt.pth --path ./DOTA_devkit/example/train_split/images/P0706__1.0__0___0.png