We now add the cpu multi process version of "ImgSplit" and "ResultMerge", gpu version of polygon nms.
The code is useful for DOTA or ODAI. The code provide the following function
- Load and image, and show the bounding box on it.
- Evaluate the result.
- Split and merge the picture and label.
Dota is a large-scale dataset for object detection in aerial images. It can be used to develop and evaluate object detectors in aerial images. We will continue to update DOTA, to grow in size and scope and to reflect evolving real-world conditions. Different from general object detectin dataset. Each instance of DOTA is labeled by an arbitrary (8 d.o.f.) quadrilateral. For the detail of DOTA-v1.0, you can refer to our paper.
DOAI2019 is a contest of Detecting Objects in Aerial Images on CVPR'2019. It is based on DOTA-v1.5.
DOAI2018 is a contest of object detetion in aerial images on ICPR'2018. It is based on DOTA-v1. The contest is closed now.
- install swig
sudo apt-get install swig
- create the c++ extension for python
swig -c++ -python polyiou.i
python setup.py build_ext --inplace
- Reading and visualizing data, you can use DOTA.py
- Evaluating the result, you can refer to the "dota_evaluation_task1.py" and "dota_evaluation_task2.py" (or "dota-v1.5_evaluation_task1.py" and "dota-v1.5_evaluation_task2.py" for DOTA-v1.5)
- Split the large image, you can refer to the "ImgSplit"
- Merging the results detected on the patches, you can refer to the ResultMerge.py
An example is shown in the demo. The subdirectory of "basepath"(which is used in "DOTA.py", "ImgSplit.py") is in the structure of
.
├── images
└── labelTxt