This paper presents a real-time monocular SLAM system that utilizes a point map and search-by-projection approach for accurate localization and mapping. The system maintains a map of 3D points observed in multiple frames and extracts keypoints and descriptors from each frame. It establishes correspondences between frames and map points using search-by-projection, recording observations based on proximity and feature similarity. Pose optimization is employed to refine camera poses and improve trajectory estimation. Experimental results demonstrate robust localization and mapping in real-time, even in challenging environments with dynamic scenes and significant camera motion. The proposed system offers efficiency for immediate feedback and response in various applications.