The founders of Imense are Dr Christopher Town and Dr David Sinclair. In their academic lives they developed the first 'Ontological Query Evaluation Language' (OQUEL) for image retrieval, which mapped a plain text user query onto a query over automatically recognized visual content in a corpus of images.
Technology derived in spirit from OQUEL is in routine use on the Imense PictureSearch portal. The user interface allows a user to type a plain text query that is probabilistically parsed to recognise visual aspects (like 'purple center green background' or 'group of five people') and non visual aspects (e.g. 'freedom' or 'Buddhism' or 'Parma ham'). The issues associated with scaling up image search to cope with tens of millions or more images were addressed with active support from the Science and Technology Facilities Council and GridPP. News articles about Imense search technology include.
The research focus of Imense Ltd remains ontology based image content recognition. Imense uses cutting edge techniques from machine learning to build and train extremely-high-dimensional classifiers to help semantically label things in the visual world. Imense appears to use different types of visual models for different object classes — for example, Bayesian Constrained Local Models for parametric face modeling, and SVMs for general semantic content labeling.
- Language-based Querying of Image Collections on the basis of an Extensible Ontology (Town and Sinclair, 2004)