MoIL

Update: Our previous work has been accepted by PerCom (WorkShop) 2024 and won the "Best WiP Paper Award". link

The implementation of the SSL baselines is mainly based on the excellent work CL-HAR link.

Quick Start

This is our proposed method MoIL (CNNRNN)

python main_SSL.py --framework 'CNNRNN' --backbone 'CNNRNN' --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 

Try baseline methods - simCLR:

python main_SSL.py --framework 'simclr'  --backbone 'CNNRNN' --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 

Baseline method - simsiam:

python main_SSL.py --framework 'simsiam' --backbone 'CNNRNN'  --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 --criterion cos_sim 

Baseline method - BYOL:

python main_SSL.py --framework 'byol' --backbone 'CNNRNN'  --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 --criterion cos_sim 

Baseline method - Bert

python main_SSL.py --framework 'SSL' --backbone 'Transformer'  --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 --criterion mse 

Baseline method - Multi task

python main_SSL.py --framework 'multi' --backbone 'CNN'  --dataset 'openpack' --user_name 'U0101' --n_epoch 1000  --batch_size 2048 --criterion binary 

Supported Datasets

• OpenPack link
• Skoda link

Download training data

|data
├──OpenPackDataset
     └── v_3.1
├──omeData
     └── raw
├──skodaData
     └── skoda_wd
├──LogiData
     └── LogiData_wd

SSL Models

Refer to ```models/frameworks.py```. For sub-modules (projectors, predictors) in the frameworks, refer to ```models/backbones.py```
- SimSiam
- BYOL
- SimCLR
- Multi-task
- Transformer (Masked reconstruction)
- MoIL (CNNRNN, Proposed)

Reference

If you find any of the codes helpful, kindly cite our paper.

>@inproceeding{xia2024preliminary,
>  title={Preliminary Investigation of SSL for Complex Work Activity Recognition in Industrial Domain via MoIL},
 > author={Xia, Qingxin and Maekawa, Takuya and Morales, Jaime and Hara, Takahiro and Oshima, Hirotomo and Fukuda, Masamitsu and Namioka, Yasuo},
 > booktitle={2024 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)},
>  pages={465--468},
 > year={2024},
>  organization={IEEE}
>}

Related Links

Part of the augmentation transformation functions are adapted from

• https://github.com/emadeldeen24/TS-TCC
• https://github.com/terryum/Data-Augmentation-For-Wearable-Sensor-Data
• https://github.com/LijieFan/AdvCL/blob/main/fr_util.py

Part of the contrastive models are adapted from

• https://github.com/Tian0426/CL-HAR
• https://github.com/lucidrains/byol-pytorch
• https://github.com/lightly-ai/lightly
• https://github.com/emadeldeen24/TS-TCC
• https://github.com/harkash/contrastive-predictive-coding-for-har

Loggers used in the repo are adapted from

• https://github.com/emadeldeen24/TS-TCC
• https://github.com/fastnlp/fitlog