Hybrid Routing Transformer for Zero-Shot Learning

Overview

This repository is the official pytorch implementation of Hybrid Routing Transformer for Zero-Shot Learning.

Requirements

Experiments were done with the following package versions for Python 3.6:

• PyTorch 1.4.0 with CUDA 10.0
• Numpy 1.16.4
• Tensorboard 1.10.0
• Tensorflow 1.2.0
• H5py 2.9.0
• Pandas 0.24.2
• Torchvision 0.5.0

Data Preparation

1. Please download and data into the ./data folder. We show the details about download links.

2. Please run the feature extraction scripts in the ./extract folder. to get the feature maps of Resnet101.

Train and Test

For different datasets (AWA2/CUB/SUN), you can run the code:

python HRT_AWA2.py 
python HRT_CUB.py 
python HRT_SUN.py