Images, Video & Computer Vision

We’ve labeled over 10 million images for some of the most ambitious computer vision projects in the world. We’re sure we can help make your next project a success.

Our robust suite of computer vision solutions can process any annotation job from simple image classification to pixel labeling semantic segmentation. Our human-machine-teaming platform seamlessly integrates expert human analysts with cutting-edge machine learning tools to create the best training data capabilities in the industry.

Our Platform Powers:

Image Transcription

Our latest image transcription tool allows annotators to label and transcribe directly inside the application, and it supports capabilities such as bounding boxes and custom ontologies. By combining image annotation and transcription in one task, annotators can localize their transcriptions and train more accurate OCR models. In addition, machine learning assistance can be enabled, which can increase transcription rates by 30% compared to manual transcription alone.

Image Classification

Figure Eight Federal’s powerful platform can categorize the content of all your images and photos quickly and with high accuracy. You simply define your specific ontology, and our human-machine-teaming capabilities take it from there. Your images can be classified by quality (detecting blurry images, for example), type (such as product vs. lifestyle images), content (specific things that the images actually contain), or any other judgments you require to categorize your library of images, no matter how large. Federal partners have used our cutting-edge capabilities for many real world applications, from geospatial data to intelligence products.

Object Detection

Computer vision projects often need in-image labels. Figure Eight Federal’s flexible platform includes object detection tools with automatic labeling, which include capabilities for bounding boxes, polygons, line labels, ellipses, and more. Moreover, all capabilities include aggregation and quality controls to make sure you get the most exacting and accurate labels possible, with simple setup and fast processing speed.

Object Tagging

Images often require tagging of multiple object classes and potentially multiple instances of each object class. Figure Eight Federal’s platform includes powerful object tagging solutions that can accomplish these tasks and more. Our human-machine-teaming capabilities allow our annotators to quickly and accurately identify multiple object classes and multiple instances within a single image, and use the ontology you create to label each instance according to your specific instructions. With fully customizable ontologies that support hundreds of classes, you can have your images labeled to your exact specifications. Federal clients have used these capabilities to automatically tag some of the most complex images available in real-world applications.

Semantic Segmentation

Figure Eight Federal’s semantic segmentation solution provides pixel-level labeling for computer vision projects that need the most detailed levels of annotation. Our solution has a robust suite of human-machine-teaming annotation tools that makes labeling easy and exact for our annotators. We also partner with expert contributor channels of highly skilled annotators who can produce the highest quality semantic segmentation labels for your computer vision models.

Machine-Learning-Assisted Video ObjectTracking

Figure Eight Federal’s video object tracking capability leverages a cutting-edge ensemble machine learning model, and patented technology, to label videos up to 100 times faster than human-only approaches. Our human annotators simply label specific objects in the first frame, and the machine learning model retains those specific annotations by following the objects as they move through the video. Annotators can then simply retain or amend the labels as necessary, instead of having to re-label each object as in standard methods.

Landmark Detection

Figure Eight Federal’s landmark tagging tool leverages dot detection to label and identify landmarks on objects. While typically used in facial recognition and robotics use cases, our dot detection and labeling tool gives users the ability to train machine learning models on key points in imagery, increasing efficiency and robustness. And as with all our capabilities, we also provide aggregation and powerful quality controls to ensure your landmark detection jobs return the best training data possible.

Landmark Tagging

Similar to our object tagging solution, and building upon our landmark detection capability, Figure Eight Federal’s landmark tagging solution powers landmark tagging of image content based on user-customizable ontologies. This capability uses our dot detection tool to identify multiple image content classes from your custom ontology and then train your data models using key points in the imagery, thereby increasing efficiency and robustness.

Figure Eight Federal can make your machine learning projects a success. Contact us and we’ll get you started.