CascadeRoiTrans/CascadeRoiTransPack

• Cascade Roi Transformer
• 作者
• 介绍
• • 1.功能
  • 2.性能
  • 3.评价指标
  • 4.使用的数据集
• 运行环境与依赖
• 输入与输出
• 运行方式
• • train
  • Test
  • 直接调用
• 引用

Cascade Roi Transformer

作者

OYLH,Guangmiao Guo,Fuyan Ma,Leyuan Fang,yangzhouxin

介绍

1.功能

• Cascade+RoiTransformer
• 旋转框检测
• Double head+CBAM

2.性能

• 火箭军比赛11/300+
• 占用资源：2080Ti batchsize:2

3.评价指标

mAP

4.使用的数据集

火箭军比赛数据集

运行环境与依赖

• Linux
• Python 3.5+ (Say goodbye to Python2)
• PyTorch 1.1
• CUDA 9.0+
• NCCL 2+
• GCC 4.9+
• mmcv

1.创建一个conda虚拟环境，然后安装Cython.

conda create -n CascadeROITrans python=3.7 -y
source activate CascadeROITrans

conda install cython

2.根据official instructions.安装PyTorch和torchvision

3.cuda extensions.

./compile.sh

4.安装

pip install -r requirements.txt
python setup.py develop
# or "pip install -e ."

输入与输出

• |输入|RGB图像
• |输出|(class_id, confidence, x1, y1, x2, y2, x3, y3, x4, y4)

运行方式

train

# single GPU
python tools/train.py ${CONFIG_FILE}
# multi GPU
bash tools/dist_train.sh ${CONFIG_FILE} ${GPU_NUM} [optional arguments]

• --validate :边训练边测试
• --work_dir ${WORK_DIR}: 保存checkpoints等路径
• --resume_from ${CHECKPOINT_FILE}: 加载checkpoint路径

Test

# single-gpu testing
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [--out ${RESULT_FILE}]

# multi-gpu testing
./tools/dist_test.sh ${CONFIG_FILE} ${CHECKPOINT_FILE} ${GPU_NUM} [--out ${RESULT_FILE}]

直接调用

python tools/test_PENGCHENG.py --config --checkpoints

引用

@article{chen2019mmdetection,
  title={MMDetection: Open mmlab detection toolbox and benchmark},
  author={Chen, Kai and Wang, Jiaqi and Pang, Jiangmiao and Cao, Yuhang and Xiong, Yu and Li, Xiaoxiao and Sun, Shuyang and Feng, Wansen and Liu, Ziwei and Xu, Jiarui and others},
  journal={arXiv preprint arXiv:1906.07155},
  year={2019}
}

@InProceedings{Ding_2019_CVPR,
author = {Ding, Jian and Xue, Nan and Long, Yang and Xia, Gui-Song and Lu, Qikai},
title = {Learning RoI Transformer for Oriented Object Detection in Aerial Images},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}