Gesture analysis has turned out to be formulate as effective means for improving the security systems and combating the crime through anticipation and identification. In this research, the primary focus demonstrates the machine learning technique for gesture recognition that quickly detects dangerous poses or bizarre actions. The contribution of computer vision techniques is used in training this model with different gesture arrays connected to aggressive and non-aggressive postures of humans. The system can identify and distinguish between positive and negative gestures, using motion tracking, body posture, and evaluation of situational contexts, and notify security when there is potential danger. The accuracy of the proposed method in identification of high-risk gestures was tested with real environment video as well as simulated environment video and thus found be quite promising. The study demonstrates that gesture analysis in public spaces enhances crime prevention and follows up on safety guidelines. In this way, the approach presented can gradually amplify the effectiveness of security systems both in business and public spaces.