I am Derrick Richard, a Computer Science and Data Science student selected for Google Summer of Code 2026 with the mlverse organization. My project focuses on extending TorchVision support in R for computer vision tasks.

The core objective is to bring full-featured computer vision capabilities to the R ecosystem by implementing item transforms for detection and segmentation pipelines, adding state-of-the-art object detection models like RF-DETR, and building utility functions for rotated bounding box operations.

During the program, I contributed 21 pull requests, including item transforms for detection & segmentation pipelines, the RF-DETR object detection model, SAHI, rotated bounding box utilities, and a CUDA roi_align_rotated op — with 20 merged.

Overall, I've contributed 49 pull requests to the mlverse ecosystem across mlverse/torchvision and mlverse/torchvisionlib — spanning dataset loaders, model families, image transforms, documentation, and CUDA-level operations, with 38 merged.

My mentors and the mlverse community have been incredible. This experience has deepened my understanding of computer vision internals, R package development, and collaborative open-source engineering.