The overall training consists of three steps: training of the baseline model (train_baseline.py), shift prediction model (train_SPM.py), and unsupervised domain adaption (train_UDA.py). Each step requires updating the yml file according to the directory of the data, processing the specified panoramic video data, and training. The example yml files (train_baseline.yml, train_SPM.yml, and train_UDA.yml) corresponding to the three training steps are stored in the config folder.
conda create -n SUQE python=3.7 -y && conda activate SUQE
pip install -r requirements.txt
1. Download the dataset through https://github.com/Archer-Tatsu/VQA-ODV.
python sequence_extraction.py
python create_lmdb.py --opt_path step1.yml
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 --master_port=12345 train_step1.py --opt_path step1.yml
#CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --nproc_per_node=2 --master_port=12354 train_step1 --opt_path step1.yml
#CUDA_VISIBLE_DEVICES=0 python train_step1 --opt_path step1.yml
python PanEnh.py
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent