Object Heading Detection

Abstract

OHDet can be applied to rotation detection and object heading detection. Its structure combines many of my previous research contents, including R³Det, IoU-Smooth L1 Loss, CSL, etc.
Project page at https://yangxue0827.github.io/CSL_GCL_OHDet.html

We also recommend a tensorflow-based rotation detection benchmark, which is led by YangXue.

Pipeline

The figure below is the architecture of the proposed detector (RetinaNet as an embodiment).

Latest Performance

OHD-SJTU-L

Model	Model Link	PL	SH	SV	LV	HA	HC	AP50	AP75	AP50:95	Configs
R2CNN	-	90.02	80.83	63.07	64.16	66.36	55.94	70.06	32.70	35.44	-
RRPN	-	89.55	82.60	57.36	72.26	63.01	45.27	68.34	22.03	31.12	-
RetinaNet-H		90.22	80.04	63.32	63.49	63.73	53.77	69.10	35.90	36.89	cfgs_res101_ohd-sjtu-all_v1.py
RetinaNet-R		90.00	86.90	63.24	86.90	62.85	52.35	72.78	40.13	40.58	cfgs_res101_ohd-sjtu-all_v2.py
R3Det		89.89	87.69	65.20	78.95	57.06	53.50	72.05	36.51	38.57	cfgs_res101_ohd-sjtu-all_r3det_v1.py
OHDet (ours)		89.73	86.63	61.37	78.80	63.76	54.62	72.49	43.60	41.29	cfgs_res101_ohd-sjtu-all_r3det_csl_v1.py

Visualization

My Development Environment

docker images: docker pull yangxue2docker/yx-tf-det:tensorflow1.13.1-cuda10-gpu-py3
1、python3.5 (anaconda recommend)
2、cuda 10.0
3、opencv(cv2)
4、tfplot 0.2.0 (optional)
5、tensorflow-gpu 1.13

Download Model

Pretrain weights

1、Please download resnet50_v1, resnet101_v1, resnet152_v1, efficientnet, mobilenet_v2 pre-trained models on Imagenet, put it to data/pretrained_weights.
2、(Recommend in this repo) Or you can choose to use a better backbone (resnet_v1d), refer to gluon2TF.

• Baidu Drive, password: 5ht9.
• Google Drive

Compile

cd $PATH_ROOT/libs/box_utils/cython_utils
python setup.py build_ext --inplace (or make)

cd $PATH_ROOT/libs/box_utils/
python setup.py build_ext --inplace

cd $PATH_ROOT/eval_devkit
sudo apt-get install swig
swig -c++ -python polyiou.i
python setup.py build_ext --inplace

Train

1、If you want to train your own data, please note:

(1) Modify parameters (such as CLASS_NUM, DATASET_NAME, VERSION, etc.) in $PATH_ROOT/libs/configs/cfgs.py
(2) Add category information in $PATH_ROOT/libs/label_name_dict/lable_dict.py     
(3) Add data_name to $PATH_ROOT/data/io/read_tfrecord_multi_gpu_ohdet.py  

2、Make tfrecord
For OHD-SJTU dataset:

cd $PATH_ROOT/data/io/OHD-SJTU
python data_crop.py

cd $PATH_ROOT/data/io/  
python convert_data_to_tfrecord.py --VOC_dir='/PATH/TO/DOTA/' 
                                   --xml_dir='labeltxt'
                                   --image_dir='images'
                                   --save_name='train' 
                                   --img_format='.png' 
                                   --dataset='OHD-SJTU'

3、Multi-gpu train

cd $PATH_ROOT/tools
python multi_gpu_train_r3det_csl_ohdet.py

Test

cd $PATH_ROOT/tools
python test_dota_r3det_csl_ohdet.py --test_dir='/PATH/TO/IMAGES/'  
                                    --gpus=0,1,2,3,4,5,6,7  

cd $PATH_ROOT/eval_devkit
python OHD_SJTU_evaluation_OHD.py

Notice: In order to set the breakpoint conveniently, the read and write mode of the file is' a+'. If the model of the same #VERSION needs to be tested again, the original test results need to be deleted.

Tensorboard

cd $PATH_ROOT/output/summary
tensorboard --logdir=.

Citation

If this is useful for your research, please consider cite.

@article{yang2020on,
    title={On the Arbitrary-Oriented Object Detection: Classification based Approaches Revisited},
    author={Yang, Xue and Yan, Junchi and He, Tao},
    year={2020}
}

@article{yang2020arbitrary,
    title={Arbitrary-Oriented Object Detection with Circular Smooth Label},
    author={Yang, Xue and Yan, Junchi},
    journal={European Conference on Computer Vision (ECCV)},
    year={2020}
    organization={Springer}
}

@article{yang2019r3det,
    title={R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object},
    author={Yang, Xue and Liu, Qingqing and Yan, Junchi and Li, Ang and Zhang, Zhiqiang and Yu, Gang},
    journal={arXiv preprint arXiv:1908.05612},
    year={2019}
}

@inproceedings{xia2018dota,
    title={DOTA: A large-scale dataset for object detection in aerial images},
    author={Xia, Gui-Song and Bai, Xiang and Ding, Jian and Zhu, Zhen and Belongie, Serge and Luo, Jiebo and Datcu, Mihai and Pelillo, Marcello and Zhang, Liangpei},
    booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    pages={3974--3983},
    year={2018}
}

Reference

• https://github.com/endernewton/tf-faster-rcnn
• https://github.com/zengarden/light_head_rcnn
• https://github.com/tensorflow/models/tree/master/research/object_detection
• https://github.com/fizyr/keras-retinanet