Clarifai expands AI product portfolio

Clarifai has announced a set of new offerings based on its deep learning AI platform for managing unstructured image, video, text and audio data.

"We have seen accelerated growth in Q1 as more and more enterprise and public sector organizations adopt AI," said Dr. Matt Zeiler, founder and CEO of Clarifai.

Clarifai's new product line-up includes the following:

Scribe Label: A data labelling platform or fully managed data labelling service that uses AI automation to speed productivity of high-quality training datasets by 100x with 10x less data needed. Recently released are new features for video labelling, polygon-interpolation and task review for large workforces.

"Our new Scribe Label product has been a powerful addition to our platform. We built it from the ground up to use AI to accelerate model development by an order of magnitude. Auto annotations unlock the power of applying AI models to annotate large datasets using semi-supervised techniques," said Zeiler.

Spacetime Search: An AI-powered, visual search engine that automatically indexes unstructured data for deep search and management. It searches for visually similar images and custom concepts in images, video and textual data and includes an intuitive user interface for bulk operations, sorting, and filtering data. Spacetime allows for deeper search into multilingual text inputs, including custom concept search and text similarity search.

Enlight Train: A rich suite of pre-trained models used to train and produce highly accurate custom models with support for a variety of different training types such as transfer learning and deep training. Custom model training can be done from a simple portal UI or through APIs. Neural networks can be built in minutes as complexities are abstracted away, exposing only the "key" essentials.

Armada Predict: A tool that analyses unstructured image, video and text data using multiple prediction types to gain insights with fast time-to-accuracy. The tool can manage thousands of model instances simultaneously, and includes a new logo detector, weapons detector and a general text token classifier detector.

Mesh Workflows: An architecture that connects AI model "building blocks" to perform complex operations on data, express complex business logic and build solutions that target specific business needs. Mesh handles multi-modal AI workflows, automated data annotation and event triggers. New workflows just released to Mesh last week include a face detection workflow and Object Character Recognition (OCR) workflows for scene detection and document detection.

Flare: An Edge AI platform that accelerates intelligent video applications from streaming, decoding, batching and inference to on-screen display/metadata output. Flare features a secure air-gapped deployment option and high accuracy for object detection and tracking.