yonghong / fall-detection-1

Fall Detection Using SVM

Fall-Detection

This is a program to detect fall and not-fall events using SVM.

For running successfully:

1. Define the path to save the extracted features of fall and not-fall events from train videos.

2. Define the path to the train videos.

3. Define the paths to all extracted fall and not-fall features (new and old).

4. Define the path to save the trained SVM.

Now we can use the trained svm to detect fall and not-fall

5. Define the path to the trained svm.

6. Define the path to a test video.

To extract the features of fall and not fall events, press the 'z' button.

Result

In the below video we can see one output of our method. The dataset to train and test the model has been collected by Mr. Farahnezhad as we see him in the video. Moreover, he has contributed to the developement of the code.

Thanks Mr. Farahnezhad...

result.mp4