Semantic Element Consistency Learning. This is the evaluation code for SECNet using PyTorch.

This code is designed to run on Ubuntu. You will need to prepare the following:

1. Visit Link (Due to anonymous submission, link will be available soon) to download ALPRO_model.pt, SECNet_model.pt, bert-base-uncased.zip, vit_base_patch16_224.zip, and ext.zip to the root directory. Unzip the corresponding files. 
2. Download videos.pt and place it in the data/msrvtt_ret/ . 

The final directory structure should look like this:

- SECNet  
  - bert-base-uncased  
  - config_release  
  - data 
    - msrvtt  
      - txt
      - videos
  - ext
  - output
  - run_scripts
  - src
  - vit_base_patch16_224
  - ALPRO_model.pt
  - SECNet_model.pt

conda create -n SECNet python=3.7
pip install cmake==3.13.0
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia 
pip install horovod
pip install -r requirements.txt 

conda activate SECNet
cd run_scripts
bash pt_SECNet.sh

conda activate SECNet
cd run_scripts
bash inf_msrvtt_ret.sh

You can modify e2e_weights_path in config_release/msrvtt_ret.json to ALPRO_model to obtain test results for ALPRO.