Skip to content

Latest commit

 

History

History
 
 

FPN_Tensorflow

README.md

Detecting Small, Cluttered and Rotated Objects via Instance-Level Feature Denoising

Abstract

Instance-Level Feature Denoising (InLD) is an important part of our paper: SCRDet++: Detecting Small, Cluttered and Rotated Objects via Instance-Level Feature Denoising and Rotation Loss Smoothing.
This repo is based on FPN, and it is completed by YangXue.

Pipeline

5

Performance

More results and trained models are available in the MODEL_ZOO.md.

DOTA1.0 (Task2)

Model Backbone Training data Val data mAP Model Link Tricks lr schd Data Augmentation GPU Image/GPU Configs
FPN (baseline) ResNet50_v1 (600,800,1024)->800 DOTA1.0 trainval DOTA1.0 test 76.03 model No 1x No 2X Quadro RTX 8000 1 cfgs_dota1.0_res50_v2.py
FPN (memory consumption) ResNet152_v1d (600,800,1024)->MS DOTA1.0 trainval DOTA1.0 test 81.23 model ALL 2x Yes 2X Quadro RTX 8000 1 cfgs_dota1.0_res152_v1.py

Visualization

1

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 1.13

Download Model

Pretrain weights

1、Please download resnet50_v1, resnet101_v1 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, refer to gluon2TF.

Compile

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

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.py

2、Make tfrecord
For DOTA dataset:

cd $PATH_ROOT\data\io\DOTA
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='DOTA'

3、Multi-gpu train

cd $PATH_ROOT/tools
python multi_gpu_train.py

Test

cd $PATH_ROOT/tools
python test_dota_ms.py --test_dir='/PATH/TO/IMAGES/'  
                       --gpus=0,1,2,3,4,5,6,7   
                       -s (visualization, optional)
                       -ms (multi-scale test, optional)

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=.

3

5

6

4

Citation

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

@article{yang2020scrdet++,
    title={SCRDet++: Detecting Small, Cluttered and Rotated Objects via Instance-Level Feature Denoising and Rotation Loss Smoothing},
    author={Yang, Xue and Yan, Junchi and Yang, Xiaokang and Tang, Jin and Liao, Wenglong and He, Tao},
    journal={arXiv preprint arXiv:2004.13316},
    year={2020}
}

Reference

1、https://github.com/endernewton/tf-faster-rcnn
2、https://github.com/zengarden/light_head_rcnn
3、https://github.com/tensorflow/models/tree/master/research/object_detection
4、https://github.com/fizyr/keras-retinanet