• 4 min read
Apple's M1 is up to 3.6x as fast at training machine learning models
We compared the Apple M1 chip to the Intel Core i5 chip on an object detection task using Create ML.
• 2 min read
Unveiling Our New Limited Edition T-Shirt
A huge benefit of working for Roboflow is interacting with all the builders and creators using our platform. Every day, Roboflow users are experimenting with computer vision to solve fascinating problems and we're always impressed with their creativity.
We especially love it when we can feature their work
• 2 min read
Introducing Roboflow Annotate
Since we launched Roboflow in early 2020, our vision has always been to improve and streamline the workflow of computer vision projects so that developers can focus on the parts of their project that are unique, not on reinventing the wheel. But there's been one part of the
• 1 min read
Share Your Datasets with the Computer Vision Community
Computer vision problems start with finding high quality image datasets. Fortunately, access to common image data is increasingly easier. Datasets like Microsoft's COCO dataset and the Pascal VOC dataset provide a standard for common objects and measuring the efficacy of state-of-the-art computer vision models (like Scaled-YOLOv4, PP-YOLO, YOLOv4,
• 1 min read
Roboflow Changelog: December 2020
Welcome to the first of our monthly changelogs, where we will be cataloging our recent feature additions and improvements.
Roboflow Organize
* Added ability to rebalance train/test split
* Clarified pro plan features in-app
* Sped up thumbnail generation
* Added image upload by URL to the upload API
* Paginated the image listing
• 8 min read
Scaled-YOLOv4 is Now the Best Model for Object Detection
(based on Microsoft COCO benchmarks)
The object detection space remains white hot with the recent publication of Scaled-YOLOv4, establishing a new state of the art in object detection.
⚡RF-DETR Neural Architecture Search (NAS) is faster and more accurate than Scaled-YOLOv4. Read the blog post here.
Looking to train a Scaled-YOLOv4
• 5 min read
How to Run Jupyter Notebooks on an Apple M1 Mac
Learn how to run Jupyter Notebooks on Apple M1 Macbooks.
• 2 min read
Generating Renaissance Art with Computer Vision
The below post is a guest post written by Samay Lakhani and Sujay Sundar, two budding data scientists. Samay currently interns with a Silicon Valley tech company; Sujay currently does research under the mentorship of a CUNY Staten Island neuroscience professor. Both are sophomores at Jericho High School, showing that
• 2 min read
Revamping Train, Validation, Test, Split Management
Splitting data into train, validation, and test splits is essential to building good computer vision models. Today, we are announcing in-app changes to Roboflow that make it even easier to manage your train test splits as you are working through the computer vision workflow.
• 6 min read
Google Researchers Say Underspecification is Ruining Your Model Performance. Here's Five Ways to Fix That.
We read that Google underspecification paper so you don't have to.
• 4 min read
Bringing Street Murals to Life with Computer Vision
In Bedford–Stuyvesant, Brooklyn (BedStuy), Yuri Fukuda regularly walks by a mural that showcases prominent female leaders. Since October 2005, a stunning 3,300 square foot mural, When Women Pursue Justice by ArtMakers-NYC, has had a central place in the BedStuy art scene. The mural celebrates 90 women who have
• 5 min read
YOLOv4 - Ten Tactics to Build a Better Model
The YOLO v4 repository is currently one of the best places to train a custom object detector, and the capabilities of the Darknet repository are vast. In this post, we discuss and implement ten advanced tactics in YOLO v4 so you can build the best object detection model from your custom dataset.
• 3 min read
Hands on with the Roboflow Infer Web Application Interface Builder
After you train a model with Roboflow Train, you're provided with three immediate ways to use your model: a curl command, the direct URL, and an Example Web App. In this post, we'll demonstrate hands on how to use the web application (as well as how
• 4 min read
Occlusion Techniques in Computer Vision
Computer vision models learn to model a task from a training set, however, like all deep learning models, they are prone to overfit the data they have been shown, making poor inferences in the wild.
Enter Occlusion Techniques
In this post, we take an overview of occlusion techniques in computer
• 3 min read
Evaluating Object Detection Models with mAP by Class
When evaluating an object detection model in computer vision, mean average precision is the most commonly cited metric for assessing performance. Remember, mean average precision is a measure of our model's ability to correctly predict bounding boxes at some confidence level – commonly mAP@0.5 or mAP@0.
• 2 min read
How to Use the Detectron2 Model Zoo (for Object Detection)
Detectron2 is Facebook's open source library for implementing state-of-the-art computer vision techniques in PyTorch. Facebook introduced Detectron2 in October 2019 as a complete rewrite of Detectron (which was implemented in Caffe). Facebook uses Detectron2 in a wide array of their products, including Portal, and notes the framework accelerates
• 5 min read
How This Fulbright Scholar is Using Computer Vision to Protect Endangered Species
The below post is by Kasim Rafiq, a conservationist, Fulbright Scholar, and National Geographic Explorer studying at UC Santa Cruz. Kasim holds a PhD in Wildlife Ecology from Liverpool John Moores University.
Introduction
Nature underpins everything. From our global economy to our wellbeing and survival. Everything relies on nature.
Unfortunately,
• 2 min read
Train Test Split Guide and Overview
In order to ensure our models are generalizing well (rather than memorizing training data), it is best practice to create a train, test split. That is, absent rigor, our models can easily overfit to a small subset of examples we've collected. Look no further than Tesla using computer
• 6 min read
Luxonis OAK-D - Deploy a Custom Object Detection Model with Depth
We are pretty excited about the Luxonis OpenCV AI Kit (OAK-D) device at Roboflow, and we're not alone. Our excitement has naturally led us to create another tutorial on how to train and deploy a custom object detection model leveraging Roboflow and DepthAI, to the edge, with depth, faster.
• 6 min read
Behind the Design of an Augmented Reality Board Game App
During the summer of 2019, I received a Facebook message from Roboflow co-founder Brad Dwyer asking me if I wanted to design a new mobile app he was working on. His previous app, Magic Sudoku, had recently generated some buzz online and even won a Golden Kitty award from Product
• 5 min read
How to Save and Load Model Weights in Google Colab
Google Colab is Google's hosted Jupyter Notebook product that provides a free compute environment, including GPU and TPU.
Colab comes "batteries included" with many popular Python packages installed, making it a choice tool for easy model experimentation. For this reason, the Roboflow Model Library includes many
• 9 min read
Using Computer Vision to Help Deaf and Hard of Hearing Communities
The below post is a lightly edited guest post by David Lee, a data scientist using computer vision to boost tech accessibility for communities that need it. David has open sourced all materials related to this project on his GitHub. The post is also available on David's blog.
• 7 min read
An Introduction to the COCO Dataset
The computer vision research community benchmarks new models and enhancements to existing models to test model performance. Benchmarking happens using standard datasets which can be used across models. With this approach, the efficacy of various models can be compared, in general, to show how one model is more or less
• 31 min read
Software Engineering Daily Podcast
Roboflow co-founder Brad Dwyer was a guest on the Software Engineering Daily podcast. Listen on your favorite podcast app (Apple Podcasts, Spotify, Overcast, Stitcher), or see the full transcript below.
Software Engineering Daily: Roboflow: Computer Vision Models with Brad Dwyer on Apple PodcastsTraining a computer vision model is not easy.
Stay Connected