Road safety faces a grave threat from the prevalence of aggressive and inattentive driving behaviors, contributing significantly to the global rise in traffic accidents and fatalities. Identifying, tracking, and proactively addressing instances of aggressive and inattentive driving are pivotal steps toward mitigating accidents and enhancing road safety. This research project seeks to explore the potential of deep learning methods for analyzing human driving behavior, with a specific focus on identifying inattentive and aggressive driving behaviors. The ultimate goal is to develop a robust and reliable system capable of accurately recognizing and categorizing such behaviors. To achieve this objective, we harness the power of deep learning algorithm. This paper also discusses the identification and categorization of aggressive and distracted driving behaviors among human drivers, using deep learning methodologies which offer substantial promise for elevating traffic safety. The literature review uncovered several noteworthy techniques, including CNN, MTCNN, transformer-based action detection systems, and fusion models inspired by densely connected convolution networks. Comparative analysis revealed improvements in accuracy and precision through the utilization of these techniques. However, the pursuit of larger and more diverse datasets, real-time implementation capabilities, and robustness in various environmental conditions remain essential requirements. Despite these challenges, deep learning-based systems offer the potential for automatic and precise analysis of intricate driving patterns, pivotal in accident prevention and overall road safety enhancement. Future research should focus on surmounting current limitations and advancing the development of more intricate deep learning models for accurate recognition and categorization of human driving behaviors.