Images

Mutual-Channel Loss for Fine-Grained Image Classification

Mutual-Channel-Loss

Code release for The Devil is in the Channels: Mutual-Channel Loss for Fine-Grained Image Classification (TIP 2020)
DOI

Changelog

  • 2020/09/14 update the code: CUB-200-2011_ResNet18.py Training with ResNet18 (TRAINED FROM SCRATCH).
  • 2020/04/19 add the hyper-parameter fine-tune results.
  • 2020/04/18 clean the code for better understanding.

Dataset

CUB-200-2011

Requirements

  • python 3.6
  • PyTorch 1.2.0
  • torchvision

Training

  • Download datasets
  • Train: python CUB-200-2011.py, the alpha and beta are the hyper-parameters of the MC-Loss
  • Description : PyTorch CUB-200-2011 Training with VGG16 (TRAINED FROM SCRATCH).

Hyper-parameter

Loss = ce_loss + alpha_1 * L_dis + beta_1 * L_div
Mutual-Channel-Loss-1
The figure is plot by NNI.

Other versions

Other unofficial implements can be found in the following: