CTMixer

Convolution Transformer Mixer for Hyperspectral Image Classification (CTMixer)

The paper titled "Convolution Transformer Mixer for Hyperspectral Image Classification" has been accepted by GRSL in Sep 2022, which is an official implementation for CTMixer.

Pytorch

Torch: 1.7.0

Python: 3.7.3

Model

Learning Rate: 0.001

Epoch: 100

Batch Size: 100

Patch Size: $11\times11$

Encoder Layer: 1

Attention Header: 4

Optimizer: Adam (weight_decay=1e-4)

Scheduler: CosineAnnealingLR

Other

We encourage researchers to cite our latest work.

We encourage researchers to achieve different comparative experiments (papers) within a code framework, to achieve a fair comparison.

If you encounter any problems reproducing the code, please do not hesitate to contact us.

E-mail: [email protected]