Hello, I tested your code on the Indian dataset. Take 20% of the dataset, AA is very low, only 93%. Then I found that the classification result of some classes is 0.

zilongzhong / SSRN

This is a tensorflow and keras based implementation of SSRNs in the IEEE T-GRS paper "Spectral-Spatial Residual Network for Hyperspectral Image Classification: A 3-D Deep Learning Framework".

Other

212 stars 74 forks source link

Hello, I tested your code on the Indian dataset. Take 20% of the dataset, AA is very low, only 93%. Then I found that the classification result of some classes is 0. #14

Closed guozhen123 closed 5 years ago

guozhen123 commented 5 years ago

Hello, I tested your code on the Indian dataset. Take 20% of the dataset, AA is very low, only 93%. Then I found that the classification result of some classes is 0.

zilongzhong commented 5 years ago

Classification performance also depends on how many epochs you trained. For more training and testing details, please check the paper https://ieeexplore.ieee.org/abstract/document/8061020