Ask the Engineers: RF-DETR Segmentation and Creating Best-in-Class Vision Models for the Edge @Roboflow
Ask the Engineers: RF-DETR Segmentation and Creating Best-in-Class Vision Models for the Edge  @Roboflow
Uploaded October 2025 | Updated November 2026, 9 minutes ago
During this live session, we interviewed Isaac Robinson and Matvei Popov, two of the machine learning engineers from Roboflow behind RF-DETR.

They re-introduced RF-DETR, a family of real-time transformer-based vision models designed for a wide variety of domains, and discussed how the latest segmentation head of RF-DETR delivers the best mix of speed and accuracy compared to models like YOLO.

​The session also featured several community projects using Roboflow's RF-DETR for object detection, we learned how most people pronounce DETR, and addressed live audience questions around creating and using the model.

Chapters
00:00 Introduction & Welcome
02:45 What is RF-DETR?
06:55 Benchmarking: RF-DETR Segmentation vs. YOLO
18:12 Community Project: Identifying Scrap with RF-DETR
21:20 Community Project: Detecting Clothing Accessories with RF-DETR
25:22 Audience Questions: Generalizability, edge deployment, tips for training models, and more

Resources
SOTA Instance Segmentation with RF-DETR Seg (Preview)
blog.roboflow.com/rf-detr-segmentation-preview

How to Deploy RF-DETR to an NVIDIA Jetson
blog.roboflow.com/how-to-deploy-rf-detr-to-an-nvidia-jetson

Join a future live session:
roboflow.com/webinar
Ask the Engineers: RF-DETR Segmentation and Creating Best-in-Class Vision Models for the EdgeHow good is YOLOv10? | Hacking Googles new VLM, PaliGemma | Community Q&A (Jun 6)Anomaly Detection: Catch Visual Defects with AILogo Recognition and Brand Analysis with AI | Learn to Use Vision Language ModelsJoin Roboflow: EngineeringRoboflow MCP: Build Vision Apps with Claude, Codex, and MoreYOLOv8: How to Train for Object Detection on a Custom DatasetWhats New in YOLOv8 | Model Deep DiveHow to use Qwen 2.5 VL | Read Nutrition Labels (and More!) with VLMsVision AI for Every Developer | Visual Intelligence Summit 2026Basketball AI: Player Tracking, Team Detection, and Number Recognition with PythonFast Segment Anything (FastSAM) vs SAM | Is it 50x faster?
Roboflow |

Ask the Engineers: RF-DETR Segmentation and Creating Best-in-Class Vision Models for the Edge

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER