The highway contains several lanes, spacious roadways, and high traffic. Expressways convey more people than regular roadways, which is crucial to the nation's economy. A highway crash will kill many people and destroy property. On the freeway, automobiles drop objects, causing major rear-end collisions. The expressway safety detection system uses video cameras to monitor crucial areas of the highway. However, coverage is limited. This research proposes driving vehicle-based expressway tossing object detection to overcome this issue. Mobile road vehicles detect expressway-throwing items. It identifies and records all traffic occurrences in real-time. Throwing things sends an alert message to the control center. After analysis and validation, the control center alerts relevant driving vehicles and manages incidents quickly. Expressway-thrown object detection systems include video capture, video detection and processing, picture transmission, and control centers. This article discusses the throwing object detection system as a moving target recognition and tracking method. Phase correlation estimates and compensates pseudo-motion. Using standard information from the current frame's prior frames creates an acting backdrop model. The current frame's different historical frames efficiently separate the moving items from the foreground. The moving target's shape and location are refined using the two-step morphological technique. To solve data association, the Kalman filter tracks moving objects using centroid, size, and intensity distribution. SVM classifiers categorize and identify moving targets and track non-vehicle targets (throwing items) based on HOG properties. The experimental findings demonstrate that the suggested technique can reliably recognize and track moving targets and discriminate moving object features to detect thrown items.