/Yavar_hackathon

Implementing a solution for fall detection in videos, with specific emphasis on detecting falls accurately and minimizing false positives.

Primary LanguagePython

Yavar_hackathon

Yavar Internship Selection Assignment

For hugging face OWL-ViT use

token = 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

Problem Statement

People's fall detection is a critical concern due to its potentially life-threatening consequences, especially in environments such as staircases, escalators, and steps. The assignment requires implementing a solution for fall detection in videos, with specific emphasis on detecting falls accurately and minimizing false positives.

Libraries/Algorithms Used

Python Streamlit YOLO Mediapipe OWL-ViT Scikit-learn

Input

Offline video files in mp4 format containing scenes where fall detection needs to be performed.

Person detection with YOLO

Person Predicted

Pose detection with mediapipe

Pose detected

fall detection with OWL-ViT

Fall detected

USER INTERFACE

Link to Video

Evaluation Criteria

  • High Accuracy: The model should accurately detect falls in the video sequences.
  • Minimal False Positives: The solution should minimize false positive detections to avoid unnecessary alarms.

Metrics

Usage

  1. Clone the repository to your local machine:

    git clone https://github.com/your-username/yavar-internship-assignment.git```