what we can do for you

Designing Monitoring Networks - Sampling Design - GIS

Based on many years of experience in monitoring and inventorying fauna, we offer assistance in designing camera trap monitoring networks.

We design your camera trap monitoring network to precisely align with your specific monitoring goals, target species, study area characteristics, and logistical constraints. Utilizing Geographic Information System (GIS) software, we conduct thorough spatial analysis to optimize network placement. You will receive a comprehensive report, along with the spatial layers, detailing the finalized network design. Furthermore, for a deeper understanding, we offer an optional service to run sophisticated simulations. These simulations estimate the minimum sampling effort required to achieve the desired statistical precision for your subsequent data analysis and predictions.

TRAPPER DB - Data Organization - Big Data - Cloud Storage

We store, organize, and classify data collected in the field in our proprietary system TRAPPER.

It is a network database application designed for effective management of camera trap projects generating millions of animal observations. It enables team expert classification of recordings, AI-based classification, database management, data browsing and analysis, as well as sharing them in the form of interactive maps and reports. Data exported from the TRAPPER system complies with the global standard for camera trap data exchange Camtrap DP, which we co-create. We support data storage in the cloud (including Microsoft Azure, OVH, Amazon S3), as well as local deployments.

TRAPPER is an open system (GNU GPL license) that can be expanded with new functionalities.

TRAPPER AI - Data Processing - AI - Deep Learning

TRAPPER AI is a component of the TRAPPER system that enables automatic image classification using artificial intelligence.

We use the latest technologies of deep machine learning (deep learning) and image recognition (image recognition), which allow for object detection and recognition of animal species in photos and videos from camera traps. We use available solutions and also train our own specialized AI models. Object detection algorithms achieve accuracy >95% and allow for filtering out empty recordings, which can constitute up to 50% of the collected material. Our proprietary algorithms for recognizing large and medium mammal species achieve an accuracy of ~90% at this stage of work. Additionally, we classify recordings with people for automatic anonymization of such data, in accordance with GDPR guidelines.

TRAPPER AI is a highly scalable system and can operate in distributed computing environments in the cloud, using multiple GPU units simultaneously. We cooperate with experienced developers and AI specialists who are able to adapt the system to the individual needs of the client.

TRAPPER CS - Citizen-Science - Distributed Monitoring Networks

TRAPPER CS is a solution that enables the creation of distributed monitoring networks, based on the assumptions of so-called citizen science, within which volunteers, foresters, researchers, nature enthusiasts, and hunters receive access to a specially designed, simplified, and visually attractive interface of the TRAPPER system. Network participants can browse and classify camera trap recordings, as well as upload their own photos and videos. The TRAPPER CS system enables the creation of networks of any scale, in which participants can be assigned to specific areas, and can also be rewarded for their work. A pilot version of the TRAPPER CS system is currently being implemented in Sweden and will support nationwide animal monitoring conducted by the Swedish Hunters’ Association (Svenska Jägareförbundet).

TRAPPER INSIGHTS - Data Analysis - Reporting

Based on the collected and processed data, we perform a wide range of ecological and population analyses. To estimate the relative and absolute abundances and densities of selected species, we use a number of statistical models available in the literature, including (Spatial) Random Encounter Model, Distance Sampling, REST, and (Spatial) Bayesian N-Mixture Model. We analyze the sex structure and daily and seasonal activity profiles of species. We specialize in spatial analyses of camera trap data, in which we map with high resolution the distribution and interactions between species and the structural and functional diversity of the composition of entire animal communities at the landscape level. We map anthropogenic pressure in the monitored area.

In the field of data analysis methods, we cooperate with leading scientific units around the world, including Swedish University of Agricultural Sciences, University of California Berkeley, University of Freiburg, Research Institute for Nature and Forest (INBO), Zoological Society of London, University of Florence, North Carolina Museum of Natural Sciences, and many others.

We are part of the European EUROMAMMALS network. We co-create the global standard for camera trap data exchange, Camtrap DP, which is being developed within the TDWG (Biodiversity Information Standards) organization and officially supported by GBIF (Global Biodiversity Information Facility).