Poly Kernel Inception Network for Remote Sensing Detection

Introduction

This repository is the official implementation of CVPR2024 Paper "Poly Kernel Inception Network for Remote Sensing Detection".

Results and models

Pretrained models

Imagenet 300-epoch pretrained PKINet-T backbone: Download

Imagenet 300-epoch pretrained PKINet-S backbone: Download

Experiments results

DOTAv1.0

Model	mAP	Angle	Aug	Configs	Download
PKINet-T (1024,1024,200)	77.87	le90	-	pkinet-t_fpn_o-rcnn_dotav1-ss_le90	model
PKINet-S (1024,1024,200)	78.39	le90	-	pkinet-s_fpn_o-rcnn_dotav1-ss_le90	model

DOTAv1.5

Model	mAP	Angle	Aug	Configs	Download
PKINet-S (1024,1024,200)	71.47	le90	-	pkinet-s_fpn_o-rcnn_dotav15-ss_le90	model

Installation

MMRotate-PKINet depends on PyTorch, MMCV and MMDetection. Below are quick steps for installation. Please refer to Install Guide for more detailed instruction.

conda create --name openmmlab python=3.8 -y
conda activate openmmlab
conda install pytorch==1.11.0 torchvision==0.12.0 cudatoolkit=11.3 -c pytorch
pip install yapf==0.40.1
pip install -U openmim
mim install mmcv-full
mim install mmdet
mim install mmengine
git clone 
cd PKINet
mim install -v -e .

Get Started

Please see get_started.md for the basic usage of MMRotate. We provide colab tutorial, and other tutorials for:

• learn the basics
• learn the config
• customize dataset
• customize model
• useful tools

License

This project is released under the Apache 2.0 license.

Citation

@InProceedings{Cai_2024_Poly,
    author    = {Cai, Xinhao and Lai, Qiuxia and Wang, Yuwei and Wang, Wenguan and Sun, Zeren and Yao, Yazhou},
    title     = {Poly Kernel Inception Network for Remote Sensing Detection},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2024},
    pages     = {27706-27716}
}