Low-Cost CNN for Automatic Violence Recognition on Embedded System

Aim:

         The main aim of the project is to auto detect any violence action in public place and send alert notification in IoT. We use deep learning and embeddedto implement.

Abstract:

         Due to the increasing number of violence cases, there is a high demand for efficient monitoring systems, however, these systems can be susceptible to failure. Therefore, this work proposes the analysis and application of low-cost yolo techniques to automatically recognize and classify suspicious events. Thus, it is possible to alert and assist the monitoring process with a reduced deployment cost. For this purpose, a dataset with violence and non-violence actions in scenes of crowded and non-crowded environments was assembled. To demonstrate the models’ validity, a prototype was developed by using an embedded Raspberry Pi platform, able to execute a model in real-time with 4 frames-per-second of speed. In addition, a warning system was developed to recognize pre-fight behavior and anticipate violent acts, alerting security to potential situations.

       Deep Learning techniques(yolo), have shown excellent results in image and video classification. In different challenges and datasets, these structures have been performing much better than previous proposals In fact, there are three main advantages of using yolo models in intelligent monitoring systems. First, they are less affected by noise in the data. Second, they achieve higher accuracy than other methods, even sometimes greater than the human eye. Lastly, they have the ability to classify people into different orientations and postures. Moreover, they also do not require a hand crafted extractor for encoding features, as was performed before the introduction of Deep Learning

Existing System:

      The existing system is monitoring the all the activities using CCTV camera in public area. But it can’t find the violence activities, only capturing the video clips and streaming in the server.

Proposed System:

       In this project we use datasets related to violence, this dataset is trained and tested, so whenever a violence scene is detected by Maixdock controller it detect the violence and inform the traffic server by using esp8266 to connect with IoT,and any warning is sent to the server and then buzzer is sounded.