Skip to content

weihancug/Remote-Sensing-Object-Detection-with-Oriented-Bouding-Box

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Remote-Sensing-Object-Detection-with-Oriented-Bouding-Box

Some object detection codes for DOTA dataset

这里用到的数据集是DOTA数据集包含15个类别:'small-vehicle', 'plane', 'large-vehicle', 'ship', 'harbor', 'tennis-court', 'round-track-field', 'soccer-ball-field', 'baseball-diamond', 'swimming-pool', 'roundabout', 'basketball-court', 'storage-tank', 'bridge', 'helicopter'

Dota数据集实例

1 首先使用data_crop.py 讲dota数据集进行切分,可以训练的大小,例如1000x1000

2 接下来使用create_data_list.py,创建一个训练集和测试集所有文件的json文件,用于模型读取

3 模型训练: 两个参数,第一个是interpreter options: -m torch.distributed.launch --nproc_per_node = 2 第二个是:--skip-test --config-file config_path DATALOADER.2 OUTPUT_DIR output_path

-m torch.distributed.launch --nproc_per_node = 2 python train_net.py --skip-test --config-file ../configs/fcos/orientedfcos_R50_1x.yaml DATALOADER.2 OUTPUT_DIR ../training_dir/orientedfcos_R_50_FPN_1x

4 模型测试:

Releases

No releases published

Packages

No packages published

Languages