In many computer vision applications, including surveillance, medical imaging, and autonomous driving, object identification and segmentation are essential. In this research, we use the current YOLOv7 architecture to offer an improved method for real-time object recognition and segmentation. YOLOv7 is renowned for its cutting-edge speed and accuracy in real-time object detection, outperforming earlier iterations of YOLO in terms of accuracy and performance. It has been difficult to incorporate segmentation features while preserving processing speed, nevertheless. Our method maintains the high classification accuracy of YOLOv7 while adding a segmentation head to create pixel-wise masks for items that are recognized. Using well-known datasets like COCO and PASCAL VOC, we test our approach's performance in terms of segmented effectiveness (Intersection over Union, or IoU) and accuracy in detection (mean average precision, or mAP). Experimental findings show that our method preserves YOLOv7's real-time processing performance while achieving a notable increase in segmentation accuracy. This study helps close the gapseparating object detection and segmentation by providing a workable solution for situations that need precise segmentation in addition to high- speed detection.