Use Foundation Models to Train Any Vision Model Without Labeling Today we are announcing Autodistill, a new library for creating computer vision models without labeling any training data. Autodistill allows you to use the knowledge of large foundation models and transfer it to smaller models for building enterprise AI applications running in real-time or at the edge. Advancements in AI research – particularly large, multipurpose, multimodal foundation models – represent a fundamental shift in capabilities of machine learning. AI models are capable of handling an unprecedented, wide array of ..read more

Autodistill is an open-source ecosystem of tools for distilling the knowledge from large, general computer vision models (i.e. Segment Anything (SAM)) into smaller models (i.e. YOLOv8). These smaller models are more suitable for edge deployment, offering greater performance in terms of inference time and compute constraints. Autodistill takes in a folder of images relevant to your project, automatically labels them using a large, general model (called a “base model”), and uses those images to train a target model. To tell Autodistill how to label images in your project, you need to specify a ..read more

Labeling large datasets can be a time-consuming and labor-intensive task. However, with advancements in deep learning and natural language processing, it is now possible to automate the labeling process. In this blog post, we will guide you through the process of using CLIP (Contrastive Language-Image Pretraining) and Roboflow to automatically label your dataset directly within a Jupyter Notebook environment. ? You can find the completed notebook for this tutorial on Google Colab What is CLIP? CLIP, developed by OpenAI, is a cutting-edge deep learning model designed to extract visual conce ..read more

Ultralytics recently released support for the Segment Anything Model (SAM) to make it easier for users to tasks such as instance segmentation and text-to-mask predictions. Combining the power of YOLOv8 with SAM is extremely helpful In the field of computer vision, object detection and instance segmentation are crucial tasks that enable machines to understand and interact with visual data. The ability to accurately identify and isolate objects in an image has numerous practical applications, from autonomous vehicles to medical imaging. In this blog post, we will explore how to convert bounding ..read more

On June 5th, 2023, at WWDC, Apple announced their biggest combination of hardware and software in years. The “one more thing” announcement this year – wherein Apple announces a product toward the end of a keynote – was the Apple Vision Pro, a new spatial computing headset integrated with the Apple ecosystem. Apple Vision Pro’s vertically integrated hardware and software platform is bringing spatial computing, what previously might have been referred to as augmented reality (AR) or mixed reality (MR), to market in 2024. The release is packed with new and never before seen hardware advancements ..read more

Introduction In this post we will walk through the process of deploying a YOLOv8 model (ONNX format) to an Amazon SageMaker endpoint for serving inference requests, leveraging OpenVino as the ONNX execution provider. We will start by setting up an Amazon SageMaker Studio domain and user profile, followed by a step-by-step notebook walkthrough. By the end of this tutorial, you will have a better understanding of how to utilize Amazon SageMaker for deploying your ONNX models. Full notebook here. Let's get started! Set up Amazon StageMaker Studio Before we dive into the notebook walkthrough, ens ..read more

Counting objects in a zone has myriad applications in computer vision. You can count the people in a particular region on a camera, identify the number of screws located in a particular area on a car part, count holes in a piece of metal, and more. Supervision, an open-source library with utilities to help you build computer vision projects, features a full suite of tools for counting objects in a zone. In this guide, we’re going to show how to count objects in a zone using supervision and PolygonZone, an accompanying tool for calculating the coordinates for zones in an image or video. Withou ..read more

Ultralytics, the developers of YOLOv3 and YOLOv5, announced YOLOv8 in January 2023, their newest series of computer vision models for object detection, image segmentation, classification, and other tasks. YOLOv8 offers a developer-centric model experience with an intuitive Python package for use in training and running inference on models. In this guide, we’re going to discuss how to use the YOLOv8 classification features to train a model that classifies whether a banana is ripe or overripe. We will use a banana dataset from Roboflow Universe for use in training our model. Without further ado ..read more

In April 2023, Meta Research released DINOv2, a method of training computer vision models that uses self-supervision to teach a model image features. DINOv2 can be used for, among other tasks, classification. DINOv2 doesn't support classification out-of-the-box. You need to train a classification model using DINOv2 embeddings. In this guide, we're going to discuss how to classify images using DINOv2 embeddings and a C-Support Vector Classification (SVC) linear classification model. By the end of this guide, we'll have a model trained on the MIT Indoor Scene Recognition dataset that classifies ..read more

Whether you're working on object detection, instance segmentation, or classification tasks, having a reliable and easy-to-use computer vision model is essential. In this blog post, we'll explore how you can leverage the power of Roboflow and YOLOv8 Instance Segmentation to streamline your workflow and achieve outstanding results. Why Use YOLOv8 Instance Segmentation? YOLOv8 is a state-of-the-art object detection algorithm known for its high accuracy and real-time performance. It's particularly effective when it comes to instance segmentation, which involves identifying and delineating individ ..read more