Git Product home page Git Product logo

lgnmnet's Introduction

LGNMNet: Lite General Network And MagFace CNN for Macro- and Micro-expression Spotting from Long Videos

Updates

  • (August 12, 2022) Release training and inference code.

Introduction

We presented an efficient expression spotting framework (LGNMNet) towards the challenge of MaE and ME spotting.

datasets

CAS(ME)2 - http://fu.psych.ac.cn/CASME/cas(me)2-en.php

SAMM Long Videos - http://www2.docm.mmu.ac.uk/STAFF/M.Yap/dataset.php

Framework

Image text

images preprocess

  • landmarks through the method FaceX-Zoo

  • extracting the optical flow features denseflow

  • removing global head motion, eye masking, ROI selection

Intersection rate Labeling

Image text

macro- and micro-expression spotting

  • combine the peak detection and adjacent peak pairing methods to recognize expression intervals.
  • macro- and micro-expression spotting through shorter frame skip and the network.
  • merge predicted intervals for improve the performance.

Training

  • update later.

    sh train.sh  # Tips:Adam better than SGD

Results

Image text

Reproducing the results

  • All the folders are placed as follows and the pre-trained weights are located under weights/.

    ├── README.md
    ├── main_train_eval.py
    ├── data
    │   ├── train_dataset.py
    ├── doc
    │   ├── IRL.jpg
    │   ├── result.jpg
    │   └── Framework.png
    ├── log
    │   ├── CAS_Macro.log
    │   ├── CAS_Micro.log
    │   ├── SAMM_Macro.log
    │   └── SAMM_Micro.log
    ├── input
    │   ├── CAS.xlsx
    │   └── SAMM.xlsx
    ├── model
    │   ├── backbone
    │   └── head
    ├── output
    │   ├── LOG_CAS_Macro.txt
    │   ├── LOG_CAS_Micro.txt
    │   ├── LOG_SAMM_Macro.txt
    │   └── LOG_SAMM_Micro.txt
    ├── preprocess
    │   ├── densflow.py
    │   └── densflow_gpu.py
    ├── utils
    │   ├── utils.py
    │   ├── calculate.py
    │   └── find_result.py
    └── weights
        ├── CAS_Macro
        ├── CAS_Micro
        ├── SAMM_Macro
        └── SAMM_Micro
  • Installation of packages using pip

    docker pull bleakie/interview:tagname
  • Preprocess dataset

    # --start --end: choose the subject for processing
    # --K: the parameter which control the length of frmae skip when preprocess
    #  CAS-MAE:6,CAS-ME:18, SAMM-MAE:37, SAMM-ME:174
  • evaluation

    # evaluation
    python main_eval.py

    All the final results are placed in the folder './log/' as follows:

    ├── LOG_CAS_Macro.txt
    ├── LOG_CAS_Micro.txt
    ├── LOG_SAMM_Macro.txt
    └── LOG_SAMM_Micro.txt 

Citation

If you find this project useful for your research, please use the following BibTeX entry.

@InProceedings{Pattern rec Letter,
    author    = {Sai, Yang},
    title     = {Lite General Network And MagFace CNN for Macro- and Micro-expression Spotting from Long Videos},
    month     = {August},
    year      = {2022},
    email     = {[email protected]}
}

lgnmnet's People

Contributors

bleakie avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.