Image segmentation is a common use case of image processing. It enables a wide variety of activities, from self-driving cars to traffic systems that are capable of governing them. Most state of the art models like Mask R-CNN and other transformer-based models perform well but have very high inference times. Therefore in this paper, a novel model architecture is proposed that can run on consumer grade hardware while giving near real time image segmentation. This is done by creating masks on regions of interest proposed by a lightweight object detection algorithm that is the YOLOv4. Such an algorithm allows for image segmentation in near-real time for resource constrained environments as compared to other state of the art models that have high inference times are unsuitable for deployment in said environments.