leihao1999 / bicnn

图像细粒度分类

FGC_CUB-200-2011

Fine Grained Image Classification on CUB-200-2011

Environment

We suggest using Anaconda to create a virtual environment for this program. Visit official website or here to download the installer(Hope there is a GUI and a browser on your deep learning server machine).

Create a new virtual environment: conda create -n pytorch python=3.6

Activate the environment on MacOS/Linux: source activate pytorch

On Windows: activate pytorch

Requirements

Note: We suggest using pip instead of conda to install following requirements on Windows. The reason is that if you choose to use conda to install something like PyTorch or numpy, in order to speed up computation, another 3 packages start with mkl will also be downloaded. However, these mkl packages have conflicts with conda on Windows and you just cannot run the program.

If you're using MacOS or Linux, just ignore the note and enjoy conda~

If you want to speed up package download, you can add Tsinghua's package repository for conda:

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --set show_channel_urls yes

Visit 清华大学开源软件镜像站 for more information.

PyTorch

Visit Official Website, choose correct OS/PM/Python-version/CUDA-version to get install command. Please install both pytorch and torchvision.

If your download speed is too slow, you can also add Tsinghua's repository specially for installing pytorch:

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda install pytorch torchvision

requests

pip install requests, note that conda doesn't contain this package.

Other requirements

conda install matplotlib pillow

Train and evaluate globally

cd FGC_CUB-200-2011
source activate pytorch
python global.py