Graybox is the Cognitiv+ solution that allows Data Scientists to build Deep Learning models in a fraction of the time than with traditional data science methods, in 3 easy steps.
Cognitiv+ web-based annotator system allows the user to upload documents in many formats, make them machine-readable and annotate them through an intuitive platform.
You can upload your pdfs, word documents or even images. Our system is able to convert even documents with extensive formatting and rich layouts. The web-based annotator system supports both NLP and Image annotations.
Make annotation projects, define tasks, invite annotators in a fraction of time and monitor the progress of your project in just one intuitive environment.
An intuitive environment that it easy to be used by non-technical people, without worrying about any technical aspect except delivering their tasks.
Upload real corporate documents which are automatically being digitalized without the need to worry about reformatting, data cleansing or any other preprocessing. Our system automatically takes care of all that, so you can focus on the actual annotation process.
Extract annotations files from any project with a single click, ready to be consumed by our GrayBox or any other third-party AutoXML software solution.
With Cognitiv+ you can train your models faster using our configurable and extendable document analysis engine based on state of the art neural networks. This will give you the ability to run complex pipelines of neural networks on documents and extract and structure results.
You can choose if you want to use our own GrayBox Training Module or integrate your own training model.
Flexible Sampling: entity recognition, document classification, sentence classification, section classification – compatible for long corporate documents where the information is merely a needle in a haystack.
Standard Neural Networks Architecture Libraries: We support text classification or sequence tagging with hyperparameter optimisation to have state-of-the-art models with few quick configurations.
Word vectors for Sample (Feature) Encoding: Use Word vectors form the basis of most recent advances in natural language processing, including language models such as Word2vec, ELMo and GloVe
Evaluation: We produce clear reports on the performance of your models to help you to evaluate and understand where to focus in order to improve your model.
Once your model has been evaluated, you can attach them to a document processing pipeline or integrate it as a component of other systems within your company.
Ingest: Use our VisionAI technology to scan and ingest scanned corporate documents of different types. With our latest technology, the system has been designed to recognise the layout of corporate documents with great accuracy.
Real Documents: Send your documents (pdf, scanned pdf, doc, docx) and via API receive our pre-trained or your trained models in JSON, XML or HTML
Powerful API: Use our API to connect models for you to embed in your solution.
Scalability: Highly available and horizontally scalable, that can be deployed on any Docker-based environment or use our high-performance cloud.