DarthYOLO: Using YOLO for Real-Time Image Segmentation

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