ushasi / Siamese-spatial-Graph-Convolution-Network

Siamese graph convolutional network for content based remote sensing image retrieval

Tensorflow-Siamese graph convolutional network for content based remote sensing image retrieval

Paper | TensorFlow

• This is a simple siamese MLP network with Tensorflow
• Retrieval MAP value of 81.79(%) and precision at top-10 retrieval of 97.14(%) on the benchmark PatternNet dataset.
• Retrieval MAP value of 69.89(%) and precision at top-10 retrieval of 93.63(%) on the benchmark UC-Merced dataset.
• MLP network is built based on TensorFlow on top of the fully-connected features obtained from Graph Convolution Network (https://github.com/nsk1001/multi-label-analysis)

• If you use this model, please cite the following paper:

Paper

• The paper is also available at: Siamese graph convolutional network for content based remote sensing image retrieval

• Feel free to cite the author, if the work is any help to you:

@article{chaudhuri2019siamese,
  title={Siamese graph convolutional network for content based remote sensing image retrieval},
  author={Chaudhuri, Ushasi and Banerjee, Biplab and Bhattacharya, Avik},
  journal={Computer Vision and Image Understanding},
  volume={184},
  pages={22--30},
  year={2019},
  publisher={Elsevier}
}