A computer vision job called “human pose estimation” locates and recognises important body joints or points in pictures or movies. It depicts how a person is arranged in space. Understanding and interpreting the movement and posture of the human body is a difficult task, hence deep learning approaches like CNN and LSTM are used for the complex and hierarchical nature of human body systems. Image input is the first step of this procedure, which finds pertinent patterns and structures. Further by forecasting the locations of joints such as the wrists, elbows, and knees, it creates a skeleton model. Capturing the meticulous details of body configurations and joint positions, CNN first extracts spatial characteristics from input images. Next, LSTM is highly skilled at modelling temporal dependencies, which makes it possible to record sequential patterns in human movements over time. Having a plethora of applications including gesture recognition, healthcare, and sports analysis. The primary subject of this study is violence detection in sports. The primary goals of this ongoing research include increasing accuracy and real-time performance. In addition, the open issues are talked about for potential future studies.