Violence Detection Through Surveillance Videos Using Combination of VGG16 and LSTM

Now-a-days in India, population is increasing. So, it affects crowd and violence in metro cities. As a citizen, it is our responsibility to prevent the violence among the people. To aim this, human surveillance is generally chosen but it results in many human errors. So an intelligent and automated system is needed to achieve this. A proposed system is decided where a combination of deep learning model such as VGG16 and a Sequential Model using LSTM layer and various dense layers. The Kaggle Dataset used is Hockey dataset containing videos with different activities like running, fighting, walking, etc. The modelis giving around 95%. Accuracy and minimum loss.