Physical Violence Detection Using Key Framing

• Author(s):

  Kaustubh Santosh Abhangkar | Kaustubh Santosh Abhangkar | Vinayak Sachin Gadekar | Shantanu Bhausaheb Kohok | Devendra Santosh Pingat

• Keywords:

  Physical Violence Detection, Key Frame Extraction, Android Application, Artificial Intelligence, Machine Learning, Deep Learning, Convolutional Neural Network, CNN-LSTM Model, Computer Vision, Firebase Realtime Database, Firebase Authentication, Firebase

• Abstract:

  The Physical Violence Detection Using Key Framing App Is An AI-based Android Application Designed To Detect Violent Activities From Video Footage Using Intelligent Frame Analysis. The System Focuses On Identifying Suspicious Physical Actions Such As Fighting, Hitting, Kicking, And Aggressive Movement Patterns By Extracting Important Key Frames From Live Or Recorded Video. These Selected Frames Are Analyzed Using A Convolutional Neural Network (CNN) Or A Hybrid CNN-LSTM Model, Which Helps The System Recognize Violence-related Patterns With Better Accuracy And Reduced Processing Load. The Application Is Developed Using Android Java/XML And Integrates Firebase Authentication, Firebase Realtime Database, And Firebase Cloud Storage For Secure User Access, Real-time Data Handling, And Cloud-based Evidence Storage. When Violent Activity Is Detected, The System Automatically Saves The Detected Frames Along With Important Details Such As Timestamp And GPS Location. It Also Sends Instant Alerts To Authorized Users, Administrators, Or Security Personnel So That Quick Action Can Be Taken. Traditional Surveillance Systems Depend Heavily On Manual Monitoring, Which Can Lead To Delayed Response, Human Error, And Missed Incidents. This Project Solves That Problem By Adding Artificial Intelligence And Automation To The Surveillance Process. The Use Of Key-frame Extraction Makes The System More Efficient Because It Avoids Analyzing Every Video Frame And Instead Focuses Only On Meaningful Frames That May Contain Suspicious Activity.

---

Other Details

• Paper id:

  IJSARTV12I6105724

• Published in:

  Volume: 12 Issue: 6 June 2026

• Publication Date:

  2026-06-24

Download Article