# fall-detection-two-stream-cnn

**Repository Path**: eden33s/fall-detection-two-stream-cnn

## Basic Information

- **Project Name**: fall-detection-two-stream-cnn
- **Description**: Real-time fall detection using two-stream convolutional neural net (CNN) with Motion History Image (MHI)
- **Primary Language**: Unknown
- **License**: GPL-3.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 1
- **Forks**: 0
- **Created**: 2021-06-10
- **Last Updated**: 2021-09-26

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# fall-detection-two-stream-cnn
Real-time fall detection using two-stream convolutional neural net (CNN) with Motion History Image (MHI)

This repository contains code for a real-time fall detection model using two-stream CNN. The optical flow stream is replaced with Motion History Image (MHI) to allow for real-time inference. The utils.py file contains utility code for generating the data, the train_model.py file creates and trains the model, and the fall_detection.py file contains code that runs the model with the weight in the weights folder either on the FDD dataset, a video, or your webcam. More detailed description of the model architecture, performance, as well as demo footage/pictures to come in the near future. Achieved fairly good cross-validated error rate on a subset of data generated. Currently working on acquiring more data and refining data generation technique.