Suspicious Activity Detection
Prof. Tonape Y. L.1, Shirkande Prathamesh2, Talekar Mayur3, Valekar Shubham4 Wagh Ashutosh5
1,2,3,4,5 SB Patil college of Engineering, Indapur, Computer Engineering
---------------------------------------------------------------------***--------------------------------------------------------------------
Abstract - Suspicious activity is forecasting the body part or joint locations of a person from an image or a video. This project will number detecting suspicious human activity from real- time CCTV footage using neural networks. human suspicious exertion is one of the pivotal problems in computer vision that has been studied for further than 15 times. It’s important because of the sheer number of operations which can benefit from exertion discovery. For illustration, mortal disguise estimation is used in operations including video surveillance, beast shadowing and behaviour understanding, subscribe language discovery, advanced mortal- computer commerce, and marker lower stir capturing. Low cost depth sensors have limitations like limited to inner use, and their low resolution and noisy depth information make it delicate to estimate mortal acts from depth images. Hence, we plan to use neural networks to control these problems.
Suspicious human exertion recognition from surveillance video is an active disquisition area of image processing and computer vision. Through the visual surveillance, mortal exertion can be covered in sensitive and public areas analogous as machine stations, road stations, fields, banks, shopping malls, academe and sodalities, parking lots, roads, etc. to help terrorism, theft, accidents and illegal parking, vandalism, fighting, chain snatching, offence and other suspicious exertion. It’s truly delicate to watch public places continuously, therefore an intelligent video surveillance is demanded that can cover the human activity in real- time and classify them as usual and unusual activity; and can induce an alert. mainly, of the disquisition being carried out is on images and not videos. Also, none of the papers published tries to use CNNs to descry suspicious activities.
Key Words: Video Surveillance, Anomaly detection,
Machine learning, Convolutional neural networks, Image processing, Background elimination, Face detection, Person recognition.