This implementation is modeified based on MJOLNIR and SAVN.
The code has been implemented and tested on Ubuntu 18.04, python 3.6, PyTorch 0.6 and CUDA 10.1
- (Recommended) Create a virtual environment using virtualenv or conda:
virtualenv ZSON --python=python3.6
source ZSON/bin/activate
conda create -n ZSON python=3.6
conda activate ZSON
- Clone the repository as:
git clone https://github.com/pioneer-innovation/Zero-Shot-Object-Navigation.git
cd Zero-Shot-Object-Navigation
- For the rest of dependencies, please run
pip install -r requirements.txt --ignore-installed
The offline data can be found here.
"data.zip" (~5 GB) contains everything needed for evalution. Please unzip it and put it into the Zero-Shot-Object-Navigation folder.
For training, please also download "train.zip" (~9 GB), and put all "Floorplan" folders into ./data/thor_v1_offline_data
Note: if you are not using gpu, you can remove the argument --gpu-ids 0
Evaluate our model under 18/4 class split
python main.py --eval \
--test_or_val test \
--episode_type TestValEpisode \
--load_model pretrained_models/SelfAttention_test_18_4.dat \
--model SelfAttention_test \
--gpu-ids 0 \
--zsd 1 \
--split 18/4Evaluate our model under 14/8 class split
python main.py --eval \
--test_or_val test \
--episode_type TestValEpisode \
--load_model pretrained_models/SelfAttention_test_14_8.dat \
--model SelfAttention_test \
--gpu-ids 0 \
--zsd 1 \
--split 14/8Note: the folder to save trained model should be set up before training.
Train our model under 18/4 class split
python main.py \
--title mjolnir_train \
--model SelfAttention_test \
--gpu-ids 0 \
--workers 8 \
--vis False \
--save-model-dir trained_models/SA_18_4/ \
--zsd 1 \
--partial_reward 1 \
--split 18/4Train our model under 14/8 class split
python main.py \
--title mjolnir_train \
--model SelfAttention_test \
--gpu-ids 0 \
--workers 8 \
--vis False \
--save-model-dir trained_models/SA_14_8/ \
--zsd 1 \
--partial_reward 1 \
--split 14/8
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent