AMD Workshop: Ultralytics YOLO Vision 2026
Workshop 1: Ultralytics YOLO on AMD Radeon: Fine-Tune, Deploy, Accelerate | Time: 2PM - 3PM
Build an end-to-end computer vision workflow with Ultralytics YOLO on a single AMD Radeon GPU. Participants will prepare a compact multi-class OBB dataset, transfer a pretrained detection model to an OBB task, complete a 3–5 minute fine-tuning run, validate the results, and deploy the trained model through an accelerated MIGraphX inference pipeline. The workshop focuses on practical, reproducible training and deployment workflows.
Speaker: Li Zheng, Product Application Engineering, AMD
Audience: AI developers, computer vision engineers, students, and technical enthusiasts with basic Python knowledge. Prior experience with Ultralytics, ROCm, OBB, or MIGraphX is not required.
Technical Requirements / Preparation: A computer with a web browser.
Workshop 2: Ultralytics YOLO26 on AMD Radeon: From an Ultralytics Checkpoint to an End-to-End Video AI Pipeline | Time: 3PM - 4PM
Take an Ultralytics YOLO26x checkpoint beyond predict() and into a production-style video pipeline on AMD Radeon. This workshop explains how YOLO26x is exported to ONNX, compiled with AMD MIGraphX for FP16 inference, and connected to hardware video decode, OpenCV HIP preprocessing, GPU NMS, and hardware encode. Attendees will then run the complete pipeline in Radeon Cloud and explore how Qwen3-VL extends real-time detection with scene understanding.
Speaker: Zihao Mu, Member of Technical Staff, Product Application Engineering, AMD
Audience: Ultralytics users who train, fine-tune, export, or deploy YOLO models and computer vision engineers moving from notebook inference to continuous video pipelines
Recommended experience level: Intermediate. Attendees should be comfortable with Python, basic YOLO inference, and Jupyter notebooks.
Technical Requirements / Preparation: Attendees only need to bring a laptop capable of running a modern web browser. All experiments will run in a preconfigured Radeon Cloud environment provided for the workshop.