Dongxu Li, Junnan Li, Hongdong Li, Juan Carlos Niebles, Steven C.H. Hoi

Official PyTorch code for ALPRO. This repository supports pre-training as well as finetuning on

• Text-Video Retrieval on MSRVTT and DiDeMo.
• Video Question Anwsering on MSRVTT and MSVD.

Our implementation is tested on Ubuntu 20.04.1 with NVIDIA A100 GPUs. Supports for other platforms and hardwares are possible with no warrant. To install the required packages:

cd env && bash install_pkg.sh

1. Download Annotations and Pre-trained Checkpoints

   • Text annotations
   • Checkpoints of pre-trained model and finetuned model
   • Externel resources
   • unzip data.zip, output.zip, ext.zip under ALPRO/.

2. Download raw videos of downstream datasets.

   • MSRVTT:

     • download train_val_videos.zip and test_videos.zip from e.g. here.

     • check md5sum:

       51f2394d279cf84f1642defd9a651e6f  train_val_videos.zip
       0af68454cec9d586e92805739f3911d0  test_videos.zip

     • unzip all the videos into data/msrvtt_ret/videos (10k in total).

     • create the following soft link:

       ln -s data/msrvtt_ret/videos data/msrvtt_qa/videos```

   • MSVD:

     • download from official release:

       wget -nc https://www.cs.utexas.edu/users/ml/clamp/videoDescription/YouTubeClips.tar

     • check md5sum:

       9bdb20fcf14d59524a6febca9f6a8d89  YouTubeClips.tar

     • unzip all the videos to data/msvd_qa/videos (1,970 videos in total).

       mkdir data/msvd_qa/videos/ 
       tar xvf YouTubeClips.tar -C data/msvd_qa/videos --strip-components=1

   • DiDeMo:
     • Following instructions and download from the official release here;
     • unzip all the videos into data/didemo_ret/videos.
     • Note there might be a couple videos missing. See here to download. However, as they account for a small portion of training set, you may feel safe to ignore.
     • Convert all the DiDeMo videos into *.mp4 format using e.g. ffmpeg.
     • We obtained 10,463 videos following these steps (with one video 77807177@N00_5753455690_1e04ccb364 missing).

3. The directory is expected to be in the structure below:

   .
   |-config_release  # configuration files
   |-data  # text annotations and raw videos
   |---didemo_ret
   |-----txt
   |-----videos
   |---msrvtt_qa/...
   |---msrvtt_ret/...
   |---msvd_qa/...
   |-env  # scripts to install packages
   |-ext  # external resources, e.g. bert tokenizer
   |-output  # checkpoints for pre-trained/finetuned models
   |---downstreams
   |-----didemo_ret
   |-------public
   |---------ckpt # official finetuned checkpoints
   |---------log # inference log
   |---------results_test
   |-----------step_best_1_mean
   |-----msrvtt_qa/...
   |-----msrvtt_ret/...
   |-----msvd_qa/...
   |-run_scripts  # bash scripts to launch experiments
   |-src  # source code

cd run_scripts
bash inf_msrvtt_ret.sh
# {'text2video': {'r1': 33.9, 'r5': 60.7, 'r10': 73.2, 'medianR': 3.0, 'meanR': 27.404}}
bash inf_didemo_ret.sh
# {'text2video': {'r1': 35.9, 'r5': 67.5, 'r10': 78.8, 'medianR': 3.0, 'meanR': 19.125}}
bash inf_msrvtt_qa.sh
# {'ratios': {'what_ratio': [68.48, 49872], 'who_ratio': [27.99, 20385], 'how_ratio': [2.25, 1640], 'where_ratio': [0.34, 250], 'when_ratio': [0.93, 677]}, 'overall_acc': 42.12, 'what_acc': 36.05, 'who_acc': 52.24, 'how_acc': 85.67, 'where_acc': 42.8, 'when_acc': 78.88}
bash inf_msvd_qa.sh
# {'ratios': {'what_ratio': [61.93, 8150], 'who_ratio': [34.6, 4554], 'how_ratio': [2.81, 370], 'where_ratio': [0.21, 28], 'when_ratio': [0.44, 58]}, 'overall_acc': 45.91, 'what_acc': 37.02, 'who_acc': 58.59, 'how_acc': 81.62, 'where_acc': 46.43, 'when_acc': 72.41}

• To finetune on downstream tasks with the pre-trained checkpoint output/pretrain/alpro_pretrained_ckpt.pt

  cd run_scripts
  bash ft_msrvtt_ret.sh
  bash ft_didemo_ret.sh
  bash ft_msrvtt_qa.sh
  bash ft_msvd_qa.sh

  For example, with MSRVTT retrieval:

  cd ALPRO/
  
  export PYTHONPATH="$PYTHONPATH:$PWD"
  echo $PYTHONPATH
  
  CONFIG_PATH='config_release/msrvtt_ret.json'
  
  horovodrun -np 8 python src/tasks/run_video_retrieval.py \ # change -np to GPUs numbers.
      --config $CONFIG_PATH \
      --output_dir /export/home/workspace/experiments/alpro/finetune/msrvtt_ret/$(date '+%Y%m%d%H%M%S')  # change to your local path to store finetuning ckpts and logs 

• Run inference with locally-finetuned checkpoints.

   cd ALPRO/
  
   export PYTHONPATH="$PYTHONPATH:$PWD"
   echo $PYTHONPATH
  
   STEP='best'
  
   CONFIG_PATH='config_release/msrvtt_ret.json'
   OUTPUT_DIR='[INPUT_YOUR_OUTPUT_PATH_HERE]'
  
   TXT_DB='data/msrvtt_ret/txt/test.jsonl'
   IMG_DB='data/msrvtt_ret/videos'
  
   horovodrun -np 8 python src/tasks/run_video_retrieval.py \
       --do_inference 1 \
       --inference_split test \
       --inference_model_step $STEP \
       --inference_txt_db $TXT_DB \
       --inference_img_db $IMG_DB \
       --inference_batch_size 64 \
       --output_dir $OUTPUT_DIR \
       --config $CONFIG_PATH

  • OUTPUT_DIR is the path after the --output_dir option in the finetuning script.
  • $STEP is a string, which tells the script to use the checkpoint $OUTPUT_DIR/ckpt/model_step_$STEP.pt for inference.

1. Download WebVid2M and CC-3M.

   • Put WebVid2M videos under data/webvid2m;
   • 💡 we downsample webvid2m videos to 10% of the original FPS to speed-up video loading;
   • change data/cc3m/txt/cc3m.json with local image paths.

2. Training Prompter:

   cd run_scripts && bash pt_prompter.sh

3. Training video-language model:

   cd run_scripts && bash pt_alpro.sh

   If you would like to use custom prompter weight, please change teacher_weights_path in config_release/pretrain_alpro.json

4. To finetune with pre-trained checkpoints, please change e2e_weights_path in the finetuning config files, e.g. config_release/msrvtt_ret.json.

If you find ALPRO useful for your research, please consider citing:

  @inproceedings{li2021align,
    title={Align and Prompt: Video-and-Language Pre-training with Entity Prompts},
    author={Dongxu Li, Junnan Li, Hongdong Li, Juan Carlos Niebles, Steven C.H. Hoi},
    booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
    year={2022}
  }

We thank members at Salesforce Research for their helpful discussions.

The implementation of ALPRO relies on resources from ClipBERT, transformers, TimeSformer, The code is implemented using PyTorch, with multi-GPU support from Horovod and gradient-checkpoint. We thank the original authors for their open-sourcing and encourage ALPRO users to cite their works when applicable.