Roboflow creates software-as-a-service products to make building with computer vision easy. Over 500,000 developers use Roboflow to manage image data, annotate and label datasets, apply preprocessing and augmentations, convert annotation file formats, train a computer vision model in one-click, and deploy models via API or to the edge. https://roboflow.com
Roboflow的外部链接
US，Remote
Roboflow转发了
? ?? Had a lot of fun integrating Roboflow with Inductive Automation Ignition to count coffee bags on a manufacturing line. This uses Roboflow workflows to send data to ignition through the MQTT modules which ingest data from topics and place them into tags.
Roboflow转发了
Ever wanted a quick way to see the differences between models? Check out the new model differences visualizer we just added to the Workflows ecosystem! This and more new features available in latest inference release: https://lnkd.in/eZhJmbV8
Join the livestream happening now to see Piotr Skalski build this system in real-time! Youtube: https://lnkd.in/gsQUTt87
Making complex video analysis easy! ?? ?? ?? Simplifying complex video analysis was a core goal behind Supervision. With the release of 0.24.0, we've taken a major step forward. This visualization was created in just 30 minutes, from start to finish. (link to the notebook in comments) Features used: - Detection filtering with polygon zones - Robust object tracking - Customizable annotators - Line zones with in/out counters (improved in 0.24.0) - Per-class counts displayed in the top right corner (new in 0.24.0) 0.24.0 Highlights: - F1 Score (detection and segmentation models support)? - LineZone makeover + multiclass counting - Tencent NCNN support ? ?? supervision-0.24.0 release notes: https://lnkd.in/dX5t7mWq As usual, thanks to all the collaborators who worked on this release. Linas Kondrackis, Onuralp SEZER ??????, Jo?o Marcos Cardoso Ramos da Silva, Juan Cruz F., Zeel B Patel, Kader M., Eduardo Ramírez, Thibault Itart-Longueville, and a few more who I could not find on LinkedIn, for all the work you've put in! ???? #computervision #python #github #opensource
Roboflow转发了
Five new tutorial notebooks are out! ?? ?? ?? Our notebook collection continues to grow! We've recently added five new tutorials covering three models: SAM2, YOLO11, and GPT-4o. - Segment Anything 2 (SAM2) - A powerful new model from AI at Meta that can segment any object in any image or video, even those it hasn't seen before. We've added notebooks demonstrating how to use SAM2 for both image and video segmentation. - YOLO11 - The latest iteration of the popular YOLO object detection model, boasting improved speed and accuracy. New notebooks showcase how to fine-tune YOLO11 for object detection and instance segmentation on custom datasets. - GPT-4o - OpenAI's most advanced multimodal AI. Fine-tuning GPT-4o on your data is now possible, and we've provided notebooks to guide you through the process. I've also written several blog posts and streamed on YouTube about each of these models. Links are in the comments below. ???? ? ?? notebooks repository: https://lnkd.in/deQexCeS #tutorial #computervision #objectdetection #multimodal #artificialintelligence
Easy video analysis with per-class counting! Supervision 0.24.0 is out! ??
Making complex video analysis easy! ?? ?? ?? Simplifying complex video analysis was a core goal behind Supervision. With the release of 0.24.0, we've taken a major step forward. This visualization was created in just 30 minutes, from start to finish. (link to the notebook in comments) Features used: - Detection filtering with polygon zones - Robust object tracking - Customizable annotators - Line zones with in/out counters (improved in 0.24.0) - Per-class counts displayed in the top right corner (new in 0.24.0) 0.24.0 Highlights: - F1 Score (detection and segmentation models support)? - LineZone makeover + multiclass counting - Tencent NCNN support ? ?? supervision-0.24.0 release notes: https://lnkd.in/dX5t7mWq As usual, thanks to all the collaborators who worked on this release. Linas Kondrackis, Onuralp SEZER ??????, Jo?o Marcos Cardoso Ramos da Silva, Juan Cruz F., Zeel B Patel, Kader M., Eduardo Ramírez, Thibault Itart-Longueville, and a few more who I could not find on LinkedIn, for all the work you've put in! ???? #computervision #python #github #opensource
Roboflow转发了
Building AI & Robots in Public by dropping hats on to New Yorkers, Acquired Y Combinator Founder, 30 under 30
My self solving whiteboard ended up becoming an official Replit guide ? ? ? ?? https://lnkd.in/eFvq3XgZ It uses Roboflow to detect your hands and turn them into a marker or eraser. Replit lets you run code and edit it live in the browser so you can mess around with it without downloading anything. GPT tries to answer whatever you draw. You can draw on your whitboard "screen" with your finger pointer and if you make the victory / peace sign, GPT will write the answer over your fingers in blue. Clear the board by putting both hands up and shrugging. #computervision #objectdetection #ai #chrome #opensource #replit
Roboflow转发了
I benchmarked YOLO11 by fine-tuning it on the 100 datasets from the RF100 dataset. Come to Roboflow's Weekly Webinar on Tuesday morning to see the results, learn more about YOLOv11, and see all the ways you can use it across Roboflow. https://lnkd.in/gn9ratXW
Roboflow转发了
Robotics | Computer Vision | Edge MLOps | Embedded Systems | Product Designer | Rapid Prototyper
Inspired by BREAKFAST's kinetic sculptures, I decided to create my own interactive art piece. The result: a flower sculpture that responds to hand gestures. It uses YOLOv8 for gesture recognition, ESP32-controlled stepper motors for movement, and Python to tie it all together. What I like most is its potential to scale - this setup could handle many more 'flowers' without major changes. Interested in the build process? I've documented everything in a YouTube video and also open-sourced 100% of the software + hardware design [see first comment for the link]. It covers the whole journey, challenges and all. #InteractiveArt #ComputerVision Roboflow Ultralytics
Roboflow转发了
This week we had first open-source contribution to Workflows ecosystem adding Google Vision OCR block. Great to see contributions like that! Contributor: https://lnkd.in/ezqmMFUd Inference release notes: https://lnkd.in/e3Tu6Ftn