Skip to content
This is an implementation for our ACMMM 2019 paper for GZSL.
Jupyter Notebook Python Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
vision
README.md
main.py
utils.py

README.md

DSEN-for-GZSL

Shaobo Min This is an implementation for our ACMMM 2019 paper for GZSL.

Introduction

This project is a pytorch implementation of [*Domain-Specific Embedding Network for Zero-Shot Recognition *], ACMMM 2019.

Requirements

  1. Python 3.6

  2. Pytorch 0.4.0

  3. CUDA 8.0

Implementations

For more backbone supports in DSEN, please check pretrainmodels and install:

Datasets Prepare

  1. Download correspond dataset to folder your ${PATH}

  2. Creat yout dataset: e.g. for CUB in repository:

    cd ./data
    python cub.py

The examples of datasets CUB, SUN, AWA2, and aPY are already given in our repository. You should modify some path in corresponding files.

Two-stage Training

  1. Run train.py to train DSEN with fixed backbone

    e.g. for training CUB

    python train.py -a dsen -d cub -s ./chechpoints/ \
    				-b 128 --pretrained --is_fix
  2. Finetune the whole DSEN

    e.g. for training CUB

    python train.py -a dsen -d cub -s ./chechpoints/ \
    				-b 16 --alpha 0.001 --lr 0.001 \
    				--epoch 180 --resume ./checkpoints/fix.model

The official training shell for the four datasets are soon provided!
The reimplementation results and models are soon provided!

You can’t perform that action at this time.