We develop novel techniques able to recognise different lifeforms, characterise their shape, movements, sequential gestures, to the level where they will describe and compute articulated movements in real time. The research outcomes would change the paradigm of monitoring biodiversity, by giving feedback on behaviours after analysing, in real time, the streaming videos.
Our solutions in computer vision and information retrieval algorithms provide an algorithmic chain capable of: analysing shapes in sequential video-frames; and extracting vision based features used to detect movements.
Once the shape has been described and analysed, the surroundings and context can be augmented, for archival, labelling, sharing with others for further study.
We integrate latest ideas and algorithms from the Machine Learning field, applied to problems of monitoring the biodiversity in a particular setting.
Our contributions to the European Horizon 2020 Cos4Cloud Project
DynAIkon Trap listed under the inventory of Wildlabs . The Inventory is a dynamic, wiki-style discovery platform for conservation technology. It is the place to explore what technology is available for your work, how others worldwide use it, and what our conservation tech community recommends.
... Read more
To test FASTCAT-Edge, a service to build a do-it-yourself camera trap that records videos and pictures of wildlife activity, and FASTCAT-Cloud, a website service to upload and analyse nature videos and pictures coming from camera traps, Cos4Cloud organised a workshop with participants from the Catalan FELIS group on the 17th of September. The workshop was held in the natural park Sant Llorenç del Munt i l’Obac (Barcelona, Catalonia).
... Read more
Image: RPi, camera, and DynAIkon’s sensor board.
DynAikon has successfully published Motion vectors and deep neural networks for video camera traps in Elsevier’s Ecological Informatics (vol. 69).
... Read more
FASTCAT-Cloud is available under iSpot : it provides an initial analysis of your pictures or videos filtering relevant frames and recordings of wildlife activity to identify species names.
... Read more
