How to Use Roboflow Universe
How to Use Roboflow Universe You're likely on Roboflow Universe to search or explore datasets for computer and machine vision or Digital Transformation (DX); find new data to add to your own project or test and deploy models from Universe in your own applications. Roboflow Universe Overview | Roboflow Universe Announcement Fo...
A Guide for Model Production
Glossary of Common Computer Vision Terms Key Metrics mAP, Precision and Recall scores are provided for all models trained with Roboflow Train, and through the notebooks in the Roboflow Model Zoo. mAP "When evaluating an object detection model in computer vision, mean average precision is the most commonly cited metric for assessing performance." Wha...
Dataset Export: YOLOv6 Model Training
MT-YOLOv6, or "YOLOv6," is a single-stage object detection framework dedicated to industrial applications, with hardware-friendly efficient design and high performance. The new object detection model is from 美团 (Meituan Dianping). YOLOv6 benchmarking Background: YOLOv6-nano achieves 35.0 mAP on COCO val2017 dataset with 1242 FPS on T4 using TensorRT...
Dataset Health Check: Improving Your Dataset
Better models start with better datasets. In this video, we walk through common tips and tricks to improve your dataset, including using Roboflow Train and the Dataset Health Check to get there. We walk through a public dataset -- Bike Helmet Detection -- provided via Syad Salman Reza on Roboflow Universe....
Exporting Your Dataset
There are 4 Ways to Export Your Dataset on Roboflow: (1) Download to a Zip Folder (2) Download with our python package - this is the most common method, as it allows you to programmatically download your images and use them to train a model with our model zoo. (3) Download with a curl command (4) Download by Raw URL Your browser does not support HTM...
Implementing Active Learning
"Active learning is a machine learning training strategy that enables an algorithm to proactively identify subsets of training data that may most efficiently improve performance. More simply – active learning is a strategy for identifying which specific examples in our training data can best improve model performance. It's a subset human-in-the-loop...
Duplicating Projects or Datasets
We've added a feature to duplicate datasets (with annotations) in just a few clicks. Create a new project and give it a Name and Annotation Group. Hover your mouse on the three dots on the right side of the existing dataset you wish to duplicate. Click the highlighted button (three horizontal dots), and do the same for the new project we created in ...
Keypoint Annotations on Roboflow
How to Annotate a Keypoint Detection Dataset on Roboflow After generating a version of your labeled "object detection" dataset, we are ready to export the data (in YOLO v5 PyTorch format) and use the code provided in this GitHub repository to convert our bounding box annotations to keypoints. How to Annotate Keypoints on Roboflow (by Alex P., Mediu...
Pose Detection
What is Pose Detection? Pose detection is the combination of position and orientation relative to a specified object. A person standing while holding a book, with an overlayed output of the inferred detections from a human pose (keypoint) model. Source Identifying the Position of Features with Keypoint Detection Pose detection tasks are commonly acc...
Merging Projects or Datasets
We've added a feature to merge datasets (with annotations) together in just a few clicks. Hover your mouse on the three dots on the right side of one of the datasets you wish to include the merged dataset. Click the highlighted button (three horizontal dots), and select "Merge Datasets." Select the datasets you wish to merge by clicking the checkbox...
Converting a Dataset to Classification
Converting an Object Detection dataset to a Classification Dataset using Roboflow's "Isolate Objects" Advanced Pre-Processing feature. Have some data in a Roboflow object detection project that you want to use in a classification project? Here's how: Generate a dataset in the source project with the I solate Objects preprocessing step, but no augmen...
Preparing for Roboflow Universe
How to create your dataset and update your ReadMe ahead of inclusion in Roboflow Universe's Featured section . Roboflow Universe is a great place to receive project feedback and recognition from Roboflow users, and the Roboflow team. We like to select the best projects to highlight on Roboflow Universe's "Featured" section and in our weekly blog an...
Sharing a Workspace with Roboflow Support
You now invite the help@roboflow.com account to your workspace to share your data with the Roboflow team. This will allow our team to access the projects in your workspace and help with debugging and/or issue resolution. To do this, go to your Workspace Settings: Roboflow Workspace homepage UI. Accessing Workspace Settings. "Workspace Settings" can...
Model Training: YOLOv5 Video Tutorial
Train a Custom YOLOv5 Model After generating a version of your labeled object detection dataset, we are ready to train a model. Here is a tutorial on training a YOLOv5 Model with the help of Roboflow's python package for easy data management. Exporting Your Dataset Once you copy your dataset export code, continue on with the video tutorial to train...
Model Training: YOLOv4-tiny Video Tutorial
Train a Custom YOLOv4-tiny Model After generating a version of your labeled object detection dataset, we are ready to train a model. Here is a tutorial on training a YOLOv4-tiny Model with the help of Roboflow's python package for easy data management. Exporting Your Dataset There are 4 Ways to Export Your Dataset on Roboflow: (1) Download to a Zip ...
Run YOLOv5 on Your Webcam
Local Inference with YOLOv5 on a Webcam Clone the Ultralytics YOLOv5 Github repository to your computer and run detect.py after saving the model weights from your custom YOLOv5 model training. Make a note of the directory (folder) in which you clone the repository. Save the file path, if you can, for easy access later. GitHub commands cheatsheet Inc...
Model Training: YOLOv5-OBB
Train a Custom YOLOv5-OBB Model What is YOLOv5-OBB? YOLOv5 for Oriented Object Detection (Roboflow Blog) Oriented bounding boxes (OBB) are bounding boxes rotated to better fit the objects represented on an angle. Take a pill detection dataset for example. Using YOLOv5-OBB (or YOLOv5-obb) we are able to detect objects that are rotated on a given fram...
Deleting Annotation Batches
Deleting Annotation Batches on Roboflow You can delete annotation batches for images within your project: Removing Annotation Batches: Images Uploaded to the Project, but Unassigned for Labeling Removing Annotation Batches: Images Uploaded to the Project, and Within the Annotation Queue Removing Annotation Batches: Images are Labeled, Approved, and ...
Python Package for OAK Deployment
Deploying to OAK Devices with Roboflow's Python Package OAK: OpenCV AI Kit The Luxonis OAK (OpenCV AI Kit) won an award as the 2022 Edge AI and Vision Product of the Year in the Cameras and Sensors category. Follow along with our Quick Start Guide if you haven't yet trained a model with Roboflow Train. If you'd like to access Accurate Train to train...
Creating a Sample App with Universe Projects
Roboflow Universe datasets can be used for project inspiration (use the search bar to look for any terms of interest!), gathering more data for a custom project, and even testing inference on trained Universe models. Brad Dwyer, Roboflow CTO, created an example blackjack strategy web application that utilizes the Augmented Startups Playing Cards dat...
Model Training: YOLOS Video Tutorial
Train a Custom YOLOS Transformer Model This tutorial highlights the combination of a suite of best-in-class machine learning tools - Hugging Face Transformers, Roboflow, Weights and Biases, and AWS Sagemaker Serverless - to create a YOLOS object detection labeling, training, and deployment pipeline. Roboflow Model Library (Model Zoo): YOLOS The YOLO...
Deployment: Using the Roboflow Inference API
Model deployment can be a tedious process. Roboflow offers a number of easy deployment options, with model inference code snippets available for command line (cURL), Python, Javascript, Swift, Java, .NET, and more. Roboflow's Hosted Inference API Roboflow's deployment options available after training Your browser does not support HTML5 video. Infere...
Starting Your First Project
Start your first Object Detection project on Roboflow . Dataset Upload --> Dataset Generation & Model Training with Roboflow Train. Plus: Quick insight into Dataset Health Check, selecting preprocessing & augmentation steps, exporting in different annotation formats, and deploying your model to run inference. Topics Covered: Creating Y...