Team KIKIKIKI - AI4LIFE 2024 - Personal Trainer Solution (Updating...)

A video classification project that aims to classify different types of workouts or exercises performed by a person in a video. Our project got the consolation prize (4th place) in the AI4LIFE 2024 competition organized by the Da Nang University of Science and Technology (DUT).

Solution Description:

• Model: VideoMAE.
• Training is supported by Hugging Face.
• Data: You can download our data from Kaggle dataset here.
• Detail solution: will be updated soon at my blog here.

Project Structure:

• data/: contains the tsv files for training.
• scripts/: contains the scripts for dowloading model and test data.
• notebooks/: contains the notebooks for submitting the project.
• src/: contains the source code of the project (training, evaluation, inference, etc.).
• streamlit/: contains the streamlit app for the project.
• tools/: contains the tools for the project (data processing, labeling, etc.).

Team Members:

• Cao Kieu Van Manh
• Tran Dinh Minh Khoa