Implementation of State-of-the-Art Components in Object Detection for KerasCV

Previous iterations of GSoC combined with recent community contributions have helped TensorFlow catch up with the SOTA in image classification. However, object detection in TensorFlow is yet to bridge that gap. This project aims to contribute modular components used in modern object detectors to KerasCV. Moreover, some deliverables are proposed outside the main project to further enhance the usability of KerasCV for Computer Vision Engineers. Finally, the use of these modules will be depicted through the creation of YoloX-Darknet53 in a demo notebook and a keras.io tutorial.