/ResNet_Tensorflow_Tensorboard

This code is the object-oriented implementation of the Residual Network proposed in "Deep Residual Learning for Image Recognition"

Primary LanguagePython

ResNet_Tensorflow_Tensorboard

  • This code is the object-oriented implementation of the Residual Network proposed by "Deep Residual Learning for Image Recognition". They proposed several structures summarized in Table. 1 of the paper. I implemented the one with 50-layer which uses the powerful so-called bottleneck blocks.

  • This code is prepared for the MNIST data. You can simply modify it by changing the parameters and number of layers.

  • To do: Tune the value of parameters and hyper-parameters to converge faster (current values are all assigned almost arbitrarily)