Key points are not available for this paper at this time.
Usage of drones has become very popular but its fullest potential can only be realized when collaborated with other customized sensors, subsystem performing functionalities like image processing, navigation control, robotic actuation. This project helps in utilizing drone technology in an effective way to solve different issues of society by proposing a CNN based image processing model which is effective for aerial surveillance with various smart autonomous modes. The paper also discusses the requirement for employing such faster, simpler, and effective working methods in aerial surveillance by streamlining, and enhancing its operations by overcoming the various challenges of aerial imaging utilizing pose estimation. Apart from detecting pedestrian activities by using pose estimation, it is even trained to classify gender, detect weapons, fire accidents, and sparkling of roadside electric wires. It can be installed on a drone by following technical details and working states provided in later sections of this paper using master-slave protocol or it can also function independently as a Smart CCTV camera when not setup on the drone. The prototype is implemented using Raspberry Pi, Arduino connected with PI-camera, GPS Neo 6m, HC-SR04 ultrasonic sensor and the model uses the Tensor flow and OpenCV. Later the model is tested for being controlled by specially designed LABVIEW GUI and then Python GUI is programmed for controlling modes and display purposes. The model discussed in this paper also intended to deliver enhanced general protection but seriously encompasses other dimensions of crime like VAW-Violence against women and SGBV- Sexual and Gender-based Violence.
Dandamudi et al. (Sat,) studied this question.