COGNITIVE SCIENCE BASED REAL TIME THREAT WARNING SYSTEM USING DEEP LEARNING AND COMPUTER VISION

Abstract

It is critical for many countries to ensure public safety in detecting and identifying threats in a night, commercial places, border areas and public places. Humans, animals, and forests are extremely valuable components of our ecosystem but are consistently surrounded by a variety of threats. For instance, forest fires and large fire flames present immense risks as they can damage residential areas, forests, defence systems, and industries. While fires in the early stages can be identified by smoke detectors, sensors, and human assistants, these measures usually take too long and have a high false rate in detecting the fire flames, their range and size. Majority of past research in this area has focused on the use of image-level categorization and object-level detection techniques. As an X-ray and thermal security image analysis strategy, object separation can considerably improve automatic threat detection when used in conjunction with other techniques. In order to detect possible threats, the effects of introducing segmentation deep learning models into the threat detection pipeline of a large imbalanced X-ray and thermal dataset were investigated. In

our proposed system, we established a novel deep learning and computer vision- based threat detection model using transfer learning model by optimizing the

hyper-parameters, which includes the learning rate of the model. We further optimized the batch size and mini-batch gradient to improve detection and classification of these types of threats in real-time with greater accuracy and size in a dynamic environment so that the system is capable of making decisions without human assistance.