/MRCNN-for-Scene-Classification

Pytorch implementation of MRCNN for SceneClassification( Knowledge Guided Disambiguation for Large-Scale Scene Classification with Multi-Resolution CNNs)

Primary LanguagePython

MRCNN

Pytorch implementation of MRCNN for SceneClassification.

We refer to the structure in paper Knowledge Guided Disambiguation for Large-Scale Scene Classification with Multi-Resolution CNNs

Datasets and Pretrained Models

  1. datasets URL: https://pan.baidu.com/s/1j4pSD0--Elo0QyAkmSOcBg
ln -s $DOWNLOAD_PATH $MRCNN/data/
  1. pretrained models for extra network URL: https://pan.baidu.com/s/11O7FQLyYC81WX4h7-1MPvA
ln -s $DOWNLOAD_PATH $MRCNN/resnet/

3.pretrained model for MRCNN URL: https://pan.baidu.com/s/10b8coHP3RW9yX-uAPmTauA     Extract code: 0hrr

ln -s $DOWNLOAD_PATH $MRCNN/xq_model/

Train

cd $MRCNN/classification

run python train_MRCNN.py

Test

cd $MRCNN/classification

mkdir xq_result

run python val_MRCNN.py