Daisy

Artificial Intelligence for automatic identification

Daisy is a universal identification AI. It can be applied to anything: images, sounds, chemical spectra, even molecular data. It is fast, accurate and designed to be energy efficient. Classification models for training Daisy are built ethically, labour intensive pre-processing is not required.

Get in touch

Features & Benefits

Generic Pattern Matcher

Capable of identifying any pattern. Meaning that in addition to imagery, other patterns such as sounds, chemical spectra and text based material can be identified. Once the system learns about something it is able to identify it in the future.

Learns Adaptively

System can learn as it goes, with new material being incorporated into the
classification model whilst running and with low computational overhead. There is no requirement for the classification model to be rebuilt. Some Artificial General Intelligence (AGI) functionally is present.

Fast & Energy Efficient

Produces results in a fraction of the time taken by conventional deep learning neural networks and achieves comparable accuracy. All this can be realised on commodity computer hardware, clusters of graphics processing units equipped servers are not required.

Sectors which can benefit from Daisy technology include

CBIR (content based image retrieval)

Search the internet for images rather than text

Image metadata is not required as colour, texture and shape are analysed directly

Detect potentially harmful or undesirable content

Ecology

Monitor biodiversity

Inventory species

Investigate whole ecosystems, deforestation, agricultural crops and set-aside levels

Security

Detect and identify persons of interest

Automatically analyse X-ray imagery for firearms and contraband

Forensic examination of images stored on digital devices