This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ting Edge Computer Vision in the Field CONCEPT: Automatically observe, recognise and identify animals in
their natural habitat to generate semantically annotated
( e.g. recorded frame + species + ID + time + location... ) MOTIVATION: Ease tedious data collection and analysis in...
- Biological Sciences
- Environmental Demographics/Management
- Species Conservation
- (Nature Documentary Production) ??? CHALLENGE: Achieving good Species Coverage
Image Acquisition in Natural Environments
Transfer+Storage+Retrieval of Large Video Data UNMATCHED
ADVANTAGES: Computer Vision is markerless, non-invasive
and scales well to large populations. Dept of Computer Science, University of Bristol, UK | COMS30121 | Slide 4 / 11 Tagless Animal Identification | Tilo Burghardt et al. Real-Time Species Detection + Unique Feature Extraction visualisation of extracted spot patterns next to the original specimen detections Dept of Computer Science, University of Bristol, UK | COMS30121 | Slide 5 / 11 Tagless Animal Identification | Tilo Burghardt et al. Recognition System in a Nutshell object
View Full Document
- Fall '09