Latest Posts

Reducing Traffic with Computer Vision

How Transport for Cairo is Improving Commuting for Millions with Computer Vision Reducing traffic in well-planned cities where bus routes are well-mapped, subways are running on a predictable cadence, and

PP-YOLO Surpasses YOLOv4 - State of the Art Object Detection Techniques

Baidu publishes PP-YOLO and pushes the state of the art in object detection research by building on top of YOLOv3, the PaddlePaddle deep learning framework, and cutting edge computer vision research.

Ontology Management for Computer Vision

As their projects mature and dataset sizes grow, most teams wrestle with label and class management. Slicing and dicing data is more of an art than a science and you

How to Train EfficientNet - Custom Image Classification

In this tutorial, we will train state of the art EfficientNet convolutional neural network, to classify images, using a custom dataset and custom classifications. To run this tutorial on your

Advanced Augmentations in Roboflow

Roboflow Pro now supports Cutout and Mosaic. Recent research has shown there is still plenty of room to grow model performance through augmenting our training data. Roboflow has written extensively

Benchmarking the Major Cloud Vision AutoML Tools

Until now, there has been little independent research published on the performance of AutoML tools - (both relative to each other and against state of the art open source models)

Getting Started with VoTT Annotation Tool for Computer Vision

A guide on how to label your own computer vision dataset using Microsoft VoTT.

Roboflow can now ingest video!

One of the most common questions we get is "How can I use computer vision object detection models with video?" The answer is simple: you treat each frame as an

Improving Infrastructure Asset Management with Computer Vision

The below post is a lightly edited guest post from Result! Data, a Netherlands-based consultancy providing leading digital services. The Roboflow team thanks Gerard Mol (Managing Partner) and Brand Hop

Retail Store Item Detection using YOLOv5

This is a lightly edited guest post by Roboflow user Shayan Ali Bhatti, who used Roboflow to train an object detection model to identify items in grocery stores. Reposted with

How to Train a TensorFlow 2 Object Detection Model

With the recent release of the TensorFlow 2 Object Detection API, it has never been easier to train and deploy state of the art object detection models with TensorFlow leveraging

The TensorFlow 2 Object Detection Library is Here

The TensorFlow Object Detection API has been upgraded to TensorFlow 2.0. We discuss here what the new library means for computer vision developers and why we are so excited

What are Anchor Boxes in Object Detection?

Object detection models utilize anchor boxes to make bounding box predictions. In this post, we dive into the concept of anchor boxes and why they are so pivotal for modeling

Convert Supervisely Annotations in Two Minutes

In this post, we walk through how to download data from and convert annotations to YOLO Darknet format specifically, and more generally any data format or

Introducing Class Label Remapping and Omission

With Roboflow Pro, you can now remap and omit class labels within Roboflow as a preprocessing step for your dataset version. Class management is a powerful tool to get the most out of your training data and your hard earned class label annotations.

Roboflow vs Scale

A question we often get is "How is Roboflow different from Scale?" The truth is, Roboflow works great in conjunction with outsourced labeling services like Scale, LabelBox, Amazon SageMaker Ground

How a AI Acts as a Human Rights Watchdog for the Maasai People

The Maasai are an indigenous ethnic group in modern Kenya and northern Tanzania, tracing routes to the Great Rift Valley in East Africa as early as the 15th century. Roughly

Train YOLOv4-tiny on Custom Data - Lightning Fast Object Detection

YOLOv4-tiny has been released! You can use YOLOv4-tiny for much faster training and much faster detection. In this article, we will walk through how to train YOLOv4-tiny on your own

YOLOv5 New Version - Improvements And Evaluation

On June 25th, the first official version of YOLOv5 was released by Ultralytics. In this post, we will discuss the novel technologies deployed in the first YOLOv5 version and analyze

Improving Uno with Computer Vision (Plus the Dataset so You Can Too)

Uno card identification and scoring in real-time. (Credit: Adam Crawshaw)You've likely been playing Uno wrong all of your life. It's a simple game, right? Rid your hand of all

Creating BoardBoss: A Mobile Application that Improves Boggle

How Making an iOS Application Inspired Roboflow Before the Roboflow team was making tools for improving how developers apply computer vision to their problems, we were making our own computer

Why and How to Implement Random Rotate Data Augmentation

Computer vision data augmentation is a powerful way to improve the performance of our computer vision models without needing to collect additional data. We create new versions of our images

How to Train Detectron2 on Custom Object Detection Data

In this post, we will walk through how to train Detectron2 to detect custom objects in this Detectron2 Colab notebook. After reading, you will be able to train your custom

How to Convert Annotations from PASCAL VOC to YOLO Darknet

A bedrock of computer vision is having labeled data. In object detection problems, those labels define bounding box positions in a given image. As computer vision rapidly evolves, so, too,

How to Build a Custom Open Images Dataset for Object Detection

We are excited to announce integration with the Open Images Dataset and the release of two new public datasets encapsulating subdomains of the Open Images Dataset: Vehicles Object Detection and

Responding to the Controversy about YOLOv5

We appreciate the machine learning community's feedback, and we're publishing additional details on our methodology.(Note: On June 14, we've incorporated updates from YOLOv4 author Alexey Bochkovskiy, YOLOv5 author Glenn

YOLOv5 is Here: State-of-the-Art Object Detection at 140 FPS

Less than 50 days after the release YOLOv4, YOLOv5 improves accessibility for realtime object detection.June 29, YOLOv5 has released the first official version of the repository. We wrote a

How to Train YOLOv5 On a Custom Dataset

The YOLO family of object detection models grows ever stronger with the introduction of YOLOv5 by Ultralytics. In this post, we will walk through how you can train YOLOv5 to

Getting Started with CVAT - Annotation for Computer Vision

How to label your own computer vision dataset in CVAT.Labeling docks, boats, and jet skis in CVAT for our aerial maritime drone datasetIn order to use modern computer vision

Introducing New Roboflow Pricing

It's now even easier to scale up projects with Roboflow. We launched Roboflow in January with the mission of democratizing computer vision. Our thesis is simple: you shouldn't need to

Breaking Down YOLOv4

A thorough explanation of how YOLOv4 worksThe realtime object detection space remains hot and moves ever forward with the publication of YOLO v4. Relative to inference speed, YOLOv4 outperforms other

Teaching a Drone to Fly on Auto Pilot with Roboflow

An animated drone flying through a correctly identified gate. (Image provided via Victor Antony, animated by Roboflow)Drones are enabling better disaster response, greener agriculture, safer construction, and so much

Getting Started with Data Augmentation in Computer Vision

Data augmentation in computer vision is not new, but recently data augmentation has emerged on the forefront of state of the art modeling. YOLOv4, a new state of the art

A New Video Tutorial: YOLOv4 in PyTorch

We heard your feedback! More video walkthroughs. Many users report that video tutorials help round out the edges of their knowledge to get the most from Roboflow. Seeing how others

Introducing An Even Better Way to Preview Image Preprocessing and Augmentation

Knowing what preprocessing and augmentation steps to apply is hard. We've written many individual posts about the steps required to make informed resize decisions (how to resize images in image

How to Train a VGG-16 Image Classification Model on Your Own Dataset

Impatient? Jump to our VGG-16 Colab notebook. Image classification models discern what a given image contains based on the entirety of an image's content. And while they're consistently getting better,

Introducing a Thermal Infrared Dataset for Object Detection

Computer vision is performed on a wide array of imaging data: photographs, screenshots, videos. Commonly, this data is captured in similar perception to how humans see – along the visible red,

How to Train YOLOv4 on a Custom Dataset

In this tutorial, we walkthrough how to train YOLOv4 Darknet for state-of-the-art object detection on your own dataset, with varying number of classes. Train YOLOv4 on a custom dataset with

How to Use the GPU within a Docker Container

In this post, we walk through the steps required to access your machine's GPU within a Docker container. Configuring the GPU on your machine can be immensely difficult. The configuration

When to Use Contrast as a Preprocessing Step

Adding contrast to images is a simple yet powerful technique to improve our computer vision models. But why? When considering how to add contrast to images and why we add

Data Augmentation in YOLOv4

The "Secret" to YOLOv4 isn't Architecture: It's in Data PreparationThe object detection space continues to move quickly. No more than two months ago, the Google Brain team released EfficientDet for

Getting the Most Out of Roboflow Office Hours

Welcome To Our Office. Come with Questions, Please. The Roboflow team has been inspired and impressed with what our users are building on top of Roboflow. From making models that

When Should I Auto-Orient My Images?

The recommended Roboflow setting is "Auto-Orient: Enabled"When should you auto-orient your images?The short answer: almost always.When an image is captured, it contains metadata that dictates the orientation

What is Mean Average Precision (mAP) in Object Detection?

What is mean average precision? How do we calculate mAP?

Breaking Down the Technology Behind Self-Driving Cars

In May 2016, Joshua Brown died in the Tesla's first autopilot crash. The crash was attributed to the self-driving cars system not recognizing the difference between a truck and the

YOLOv3 Versus EfficientDet for State-of-the-Art Object Detection

YOLOv3 is known to be an incredibly performant, state-of-the-art model architecture: fast, accurate, and reliable. So how does the "new kid on the block," EfficientDet, compare? Without spoilers, we were

Breaking Down Roboflow's Health Check Dimension Insights

Roboflow improves datasets without any user effort. This includes dropping zero-pixel bounding boxes and cropping out-of-frame bounding boxes to be in-line with the edge of an image. Roboflow also notifies

Introducing the Roboflow Model Library

Over the past few months we've been building up a library of easy to use, open source computer vision models. We've now given them a home: the Roboflow Model Library.

The Difference Between Missing and Null Annotations

A discussion of missing versus null annotations and how VOC XML and COCO JSON handle them. Preparing data for computer vision models is a tedious task. Even assuming training images

A Thorough Breakdown of EfficientDet for Object Detection

In this post, we do a deep dive into the neural magic of EfficientDet for object detection, focusing on the model's motivation, design, and architecture. Recently, the Google Brain team

Roboflow for Students and Universities

We've seen tremendous interest in Roboflow from the research community. Faculty and students from institutions ranging from California to Malta to Taiwan have been using Roboflow to accelerate their computer

Roboflow Presents at Open Data Science Conference (ODSC) East 2020

The Open Data Science Conference (East) looked a bit different this year. While typically 6,000+ data science professionals gather in Boston for the Expo, the team at ODSC moved

How to Create a Synthetic Dataset for Computer Vision

The appeals of synthetic data are alluring: you can rapidly generate a vast amount of diverse, perfectly labeled images for very little cost and without ever leaving the comfort of your office. The good news is: it's easy to try! And we're about to show you how.

Training EfficientDet Object Detection Model with a Custom Dataset

A tutorial to train and use EfficientDet on a custom object detection task with varying number of classes YOLOv5 is Out! If you're here for EfficientDet in particular, stay for

Introducing an Improved Hard Hat Dataset for Computer Vision in Workplace Safety

In a given year, approximately 65,000 workers wearing hard hats incur head injuries in the workplace, of which over one thousand ultimately die. Workplace safety regulations exist to protect

Our First Video Tutorial: YOLOv3 in PyTorch on a Custom Dataset

We're introducing a new experiment this week: Roboflow is launching a Roboflow YouTube channel.We've been encouraged by the popularity of our computer vision tutorials. When Googling for some architectures,

How to Create to a TFRecord File for Computer Vision and Object Detection

TensorFlow expedites the machine learning process markedly. From abstracting complex linear algebra to including pre-trained models and weights, getting the most out of TensorFlow is a full-time job. However, when

Using Computer Vision to Fight Coronavirus (COVID-19)

As global coronavirus case numbers continue to climb, troubling stories of hospital shortages, deaths, and disrupted communities fill the news. Frankly, it can leave one feeling disempowered – especially when the

Releasing a New YOLOv3 Implementation in PyTorch

At Roboflow, we're constantly adapting our product to make it as easy as possible for users to create custom computer vision models on high quality data. While we have an

Introducing Image Preprocessing and Augmentation Previews

Knowing how an image preprocessing step or augmentation is going to appear before you write the code for it is essential. Is it worth it to figure out the right

How Flip Augmentation Improves Model Performance

Flipping an image (and its annotations) is a deceivingly simple technique that can improve model performance in substantial ways. Our models are learning what collection of pixels and the relationship

Introducing Bounding Box Level Augmentations

Having training data that matches the diversity of your task is paramount to the success of your models. At Roboflow, we’re committed to providing you with state-of-the-art techniques that

Getting Started with LabelImg for Labeling Object Detection Data

Accurately labeled data is essential to successful machine learning, and computer vision is no exception. In this walkthrough, we’ll demonstrate how you can use LabelImg to get started with

The Importance of Blur as an Image Augmentation Technique

When we train computer vision models, we often take ideal photos of our subjects. We line up our subject just right and curate datasets of best case lighting. But our

Training a TensorFlow Faster R-CNN Object Detection Model on Your Own Dataset

Following this tutorial, you only need to change a couple lines of code to train an object detection model to your own dataset. Computer vision is revolutionizing medical imaging. Algorithms

Why to Add Noise to Images for Machine Learning

We seek to build computer vision models that generalize to as many real world situations as we can, even when we cannot anticipate them. It's a bit of a catch-22:

How to Select the Right Computer Vision Model Architecture

The success of your machine learning model starts well before you ever begin training. Ensuring you have representative images, high quality labels, appropriate preprocessing steps, and augmentations to guard against

Introducing an Improved PlantDoc Dataset for Plant Disease Object Detection

The world population is expected to reach 9.7 billion by 2050. That’s a lot of mouths to feed. Technology is powering the next generation of yield increases. Computer

Why and How to Implement Random Crop Data Augmentation

We can’t capture a photo of what every object looks like in the real world. (Trying to find an image to prove the prior sentence is a fun paradox!

How to Win Pioneer is an online startup accelerator where companies are chosen based (partially) on weekly peer-review of progress updates. Roboflow has now been #1 on the global leaderboard for 18

Releasing an Improved Blood Count and Cell Detection (BCCD) Dataset

Computer vision is revolutionizing medical diagnoses by assisting doctors with patterns they may not have seen or identifying an error they may have overlooked. Thus, it's unsurprising one of the

Eliminating Boilerplate Code with Roboflow to Monitor Security Camera Footage

By using Roboflow, data scientist Alaa Senjab reduced his time to train a custom object detection model detecting guns in security camera footage while increasing machine learning model accuracy. Alaa's

A popular self-driving car dataset is missing labels for hundreds of pedestrians

And that's a problem that is extremely dangerous. Machine learning, the process of teaching computer algorithms to perform new tasks by example, is poised to transform industries from agriculture to

Training a TensorFlow MobileNet Object Detection Model with a Custom Dataset

Change two lines of code and have a custom trained object detection model leveraging the TensorFlow 1.5 API.

When to Use Grayscale as a Preprocessing Step

Grayscale allows our models to be more computationally efficient. So when **shouldn't** we grayscale our images?

You Might Be Resizing Your Images Incorrectly

Resizing images is a critical preprocessing step in computer vision. Principally, our machine learning models train faster on smaller images. An input image that is twice as large requires our

How to Convert Annotations from PASCAL VOC XML to COCO JSON

Convert from VOC XML to COCO JSON (or any format!) in four clicks.

Introducing Public Datasets

One of the most painstaking components of getting started with computer vision is getting access to clean, labeled data. For example, when the Roboflow team built BoardBoss, we painstakingly collected

Why Image Preprocessing and Augmentation Matters

Understanding image preprocessing and augmentation options is essential to making the most of your training data.

Getting Started with Roboflow

Roboflow eliminates boilerplate code when building object detection models. Get started with an example.

Training a YOLOv3 Object Detection Model with a Custom Dataset

A walkthrough of building chess piece object detection model, easily adapted to your own dataset.