Vision Based Fall Detection Model Using Raspberry Pi

Vision-based fall detection methods have demonstrated their efficacy in timely addressing falls, thereby mitigating fall-related injuries. This study introduces an automated vision-based system designed for fall detection, which promptly issues alerts upon detection. The system operates by processing real-time footage captured by a Pi camera used for surveillance monitoring. The proposed model employs LSTM (Long Short-Term Memory) architecture to classify detected events into three categories: Normal, Fall Warning, and Fall, achieving an average frame rate of 30 frames per second (FPS). The system's effectiveness was evaluated using the publicly available UR Fall Detection Dataset. Validation of the system yielded impressive results, with an accuracy of 99.44%, sensitivity of 99.12%, specificity of 99.12%, and precision of 99.59%. These findings suggest that the presented system could serve as a valuable tool for identifying human falls, thereby helping to prevent complications arising from fall injuries and reducing healthcare and productivity-related costs.