Detection Transformer (DETR) for Object Detection
Object detection is still one of the highest-ROI models in applied ML, powering quality inspection, inventory systems, safety monitoring, medical imaging, and autonomous systems. Fine-tuned detectors consistently outperform general-purpose vision models on domain-specific tasks, and they do it with a fraction of the compute, latency, and cost.
In this hands-on workshop, we'll fine-tune a DETR (DEtection TRansformer) model on a custom dataset and deploy it behind a simple UI. By the end, you'll have a working detector, a walkthrough of how to build your own datasets, and a reusable pipeline you can point at your next detection problem. A full end-to-end computer vision workflow built on infrastructure that scales from your laptop to a production cluster.
What we'll cover
A practical intro to DETR and why transformers changed object detection
How to build your own dataset: collection, labeling workflows, and common pitfalls
Fine-tuning DETR with Hugging Face Transformers
Orchestrating the pipeline with Flyte 2: cached data prep, GPU-aware training, and reproducible runs
Deploying the model with aUI, with a path to scaled inference
Patterns for extending to your own detection problem
What you'll leave with
A fine-tuned DETR model trained on a custom dataset
A reusable training and deployment pipeline you can adapt to your own data
The knowledge to build and label your own datasets for future projects
A portfolio-ready project and a certificate of participation
Who it's for
ML engineers and practitioners who want to move past pretrained demos and train detectors on their own data. Whether you're prototyping at work, evaluating infrastructure for a production CV use case, or building a portfolio project, you'll leave with code you can keep extending.
Hosted by Sage Elliott, AI Engineer at Union.ai.