17 stories
Roboflow is building the visual cortex for artificial intelligence. Every day, millions of developers around the globe and over half of the Fortune 100 use Roboflow to count cells to accelerate cancer research, improve construction site safety, digitize floor plans, preserve coral reef populations,…
Build a carton inspection agent in Roboflow Workflows with RF-DETR, Gemini, MQTT alerts, and webhook submissions containing JSON and image evidence.
How to diagnose rare-class failures and improve defect detection with active learning, oversampling, targeted augmentation, synthetic data, and RF-DETR evaluation in Roboflow.
Agentic computer vision goes beyond detection by combining perception, reasoning, action, and verification. Learn how to build a practical vision agent in Roboflow Workflows using RF-DETR, tracking, Gemini, structured outputs, and event-driven actions.
Learn how Sobel edge detection finds edges with two 3×3 kernels, how it compares to Canny, and how to run it on images and video in Roboflow.
Our mission is to make the world programmable, and accelerate vision AI into the world. That means building an ecosystem where developers can experiment, founders can turn prototypes into businesses, and people who know a problem firsthand can use vision AI to solve it.
We test GPT-6 Astra on object detection, segmentation, counting, visual reasoning, and video, with examples, benchmark results, and cost comparisons.
GPT-6 Astra can return polygon outlines for any object you describe. They are not pixel masks, so we benchmarked them against SAM 3 and show how to combine the two.
Canny edge detection in five steps (blur, gradients, NMS, thresholds, hysteresis), plus a Roboflow Workflow that runs it on images and video.
GPT-6 Astra has no video input, but a sequence of timestamped frames in one request gets you a timeline of events. Three demos: sports analytics, process monitoring and truck turnaround time.
Gaze tracking explained, plus a working pipeline: detect faces with SAM 3, track them with OC-SORT, and estimate gaze with L2CS-Net in Roboflow.
GPT-6 Astra runs zero-shot detection on your images and returns bounding boxes for the classes you define.
A reject decision must happen in milliseconds, cameras generate data continuously, and the line still runs when the WAN does not. This guide covers edge vision deployment for manufacturing, from hardware and model optimization to PLC integration and IT/OT controls.
Learn how to choose and validate an object detection confidence threshold using precision, recall, error costs, and real-world model behavior.
Mine low-confidence predictions from production images, review them with a VLM, then label and retrain.
Pulling bad parts by hand dates back to 1960s assembly lines. I built a local defect detector using a webcam, RF-DETR, and a desktop robot arm. Ten training photos and 0.2 seconds per frame run the full detect, decide, pick, and verify loop right on my desk.
Roboflow Auto Label runs Gemini 3.7 across the batch and returns bounding boxes for the classes you define.
Showing 17 of 17 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
