This repository contains the reimplementation for the Sign Language Translation from Instructional Videos paper.

All the scripts are located inside examples/sign_language/scripts.

Trained final models are provided in this google drive link. Link to final models

Clone this repository, create the conda environment and install Fairseq:

git clone https://github.com/jiSilverH/idlf23-aslt.git

conda env create -f ./examples/sign_language/environment.yml
conda activate slt-how2sign-wicv2023

pip install --editable .

python setup.py build_ext --inplace

import nltk
nltk.download('punkt')

The execution of scripts is managed with Task. Please follow the installation instructions in the official documentation. We recommend using the following

sh -c "$(curl --location https://taskfile.dev/install.sh)" -- -d -b path-to-env/slt-how2sign-wicv2023/bin

Environment issues

AttributeError: module 'importlib_resources' has no attribute 'is_resource'

• Referring to this link, pip install --upgrade hydra-core omegaconf can solve the problem.
• Ignore conflicts from other packages. For testing environment, you should not upgrade aforementioned two packages. We recommend you to make a new environment for the inference.

ImportError: numpy.core.multiarray failed to import

• Referring to this link, the error occurs because of incomplete numpy installation. Reinstall numpy with pip.

The I3D keypoints and .tsv are in the dataverse. Once you have them, they should follow this structure:

├── data/
│   └── how2sign/
│       ├── i3d_features/
│       │   ├── cvpr23.fairseq.i3d.test.how2sign.tsv
│       │   ├── cvpr23.fairseq.i3d.train.how2sign.tsv
│       │   ├── cvpr23.fairseq.i3d.val.how2sign.tsv
│       │   ├── train/
│       │   │   ├── --7E2sU6zP4_10-5-rgb_front.npy
│       │   │   ├── --7E2sU6zP4_11-5-rgb_front.npy
│       │   │   └── ...
│       │   ├── val/
│       │   │   ├── -d5dN54tH2E_0-1-rgb_front.npy
│       │   │   ├── -d5dN54tH2E_1-1-rgb_front.npy
│       │   │   └── ...
│       │   └── test/
│       │       ├── -fZc293MpJk_0-1-rgb_front.npy
│       │       ├── -fZc293MpJk_1-1-rgb_front.npy
│       │       └── ...
│       └── vocab/
│           ├── cvpr23.train.how2sign.unigram7000_lowercased.model 
│           ├── cvpr23.train.how2sign.unigram7000_lowercased.txt
│           └── cvpr23.train.how2sign.unigram7000_lowercased.vocab
└── final_models/
    └── baseline_6_3_dp03_wd_2/
        ├── ckpts
            └── checkpoint.best_reduced_sacrebleu_3.5401.pt 
        ├── generates
        └── hydra_outputs

Each of the folder partitions contain the corresponding I3D features in .npy files, provided by previous work, that correspond to each How2Sign sentence.
In addition, we provide the .tsv files for all the partitions that contains the metadata about each of the sentences, such as translations, path to .npy file, duration. Notice that you might need to manually change the path of the signs_file column.

The training data is big nearly 11GB. Therefore, we recommend you to divide the training data into 4 chunks using the below commands and then concatenate 4 chunks to make the full training dataset.

curl --range 0-2000000000 -o part1 https://dataverse.csuc.cat/api/access/datafile/51543?gbrecs=true
curl --range 2000000001-4000000000 -o part2 https://dataverse.csuc.cat/api/access/datafile/51543?gbrecs=true
curl --range 4000000001-6000000000 -o part3 https://dataverse.csuc.cat/api/access/datafile/51543?gbrecs=true
curl --range 6000000001- -o part4 https://dataverse.csuc.cat/api/access/datafile/51543?gbrecs=true
cat part1 part2 part3 part4 > train.zip

As per fairseq documentation, we work with config files that can be found in CONFIG_DIR = FAIRSEQ_ROOT/examples/sign_language/config/wicv_cvpr23/i3d_best. Select the name of the .yaml files as the experiment name desired. For the final model, select baseline_6_3_dp03_wd_2. As EXPERIMENT_NAME and run:

export EXPERIMENT=baseline_6_3_dp03_wd_2
task train_slt

Remember to have a GPU available and the environment activated. Previously to the call of the function, the .env should be updated with the following variables:

DATA_DIR: path/to/i3d/folders
WANDB_ENTITY: name/team/WANDB
WANDB_PROJECT: name_project_WANDB
NUM_GPUS: 1
CONFIG_DIR: FAIRSEQ_ROOT/examples/sign_language/config/i3d_best

task generate

Similarly to other tasks, the .env should be updated:

EXPERIMENT: EXPERIMENT_NAME
CKPT: name_checkpoint, for example: checkpoint.best_sacrebleu_9.2101.pt
SUBSET: cvpr23.fairseq.i3d.test.how2sign
SPM_MODEL: path/to/cvpr23.train.how2sign.unigram7000_lowercased.model

The task generate generates a folder in the output file called generates/partition with a checkpoint.out file that contains both the generations and the metrics for the partition. Script python scripts/analyze_fairseq_generate.py analizes raw data and outputs final BLEU and rBLEU scores, call it after the task generate in the following manner:

python scripts/analyze_fairseq_generate.py --generates-dir path/to/generates --vocab-dir path/to/vocab --experiment baseline_6_3_dp03_wd_2 --partition test --checkpoint checkpoint_best

Original paper implementation repo
Laia Tarrés, Gerard I. Gallego, Amanda Duarte, Jordi Torres and Xavier Giró-i-Nieto. "Sign Language Translation from Instructional Videos", WCVPR 2023.

@InProceedings{slt-how2sign-wicv2023,
author = {Laia Tarrés and Gerard I. Gállego and Amanda Duarte and Jordi Torres and Xavier Giró-i-Nieto},
title = {Sign Language Translation from Instructional Videos},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) :Workshops},
year = {2023}
}

• Some scripts from this repository use the GNU Parallel software.

  Tange, Ole. (2022). GNU Parallel 20220722 ('Roe vs Wade'). Zenodo. https://doi.org/10.5281/zenodo.6891516

Check the original Fairseq README to learn how to use this toolkit.