Students 2025

Thomas Luypaert (@LuypaertThomas)

I am Thomas, a tropical ecologist, and current postdoctoral researcher at the University of Life Sciences. My research lies at the interface of tropical ecology and big data science, and has the overarching aim of alleviating barriers that hold back large-scale biodiversity monitoring in tropical rainforest ecosystems. During my PhD, I developed an analytical pipeline for tracking biodiversity changes in rainforests using large acoustic datasets, without needing species-level identification. Currently, my research explores ways to extract ecologically important information from eDNA, camera traps, acoustic recordings, and visual surveys collected during expeditions across the Brazilian Amazon.

Project: As part of the Amazon Biodiversity & Carbon Expeditions (abc-expeditions.com), our team has collected approximately 45,000 camera trap videos capturing large vertebrates throughout the Brazilian Amazon. Through the CV4Ecology project, I hope to identify a workable machine learning model to detect animals and classify the most commonly observed species. Our goal is to create a scalable, generalizable algorithm that reduces the time spent sorting non-target footage (e.g., blank images) while accurately identifying common species, without the need for regional retraining.

Brett Ford

I’m a developer with the North Country Wild project at St. Lawrence University. I work as a bioinformatics scientist during the day and volunteer my time to the North Country Wild project to help manage and process camera trap and passive acoustic monitoring data. I completed my MSc at the University of British Columbia and my BS at St. Lawrence University. I’m excited about combining my experience in programming and data science with my background in ecology and evolution to help inform biodiversity monitoring efforts.

Project: Maintaining connectivity between large swaths of natural habitat is essential for the maintenance of biodiversity, however we still lack a fundamental understanding of what species live where and how animals use the landscape. As part of the Algonquin to Adirondack (A2A) Collaborative, the North Country Wild project has deployed camera traps and passive acoustic monitoring devices to understand what species live within the A2A corridor and what habitats they most often occupy. While it’s relatively easy to solicit classification of game camera images through citizen science platforms like Zooniverse, acoustic monitoring files are more difficult to sift through and classify, needing more specialized knowledge of animal sounds and a longer amount of analysis time. Therefore, I am hoping to use recent advances in computer vision to help filter through the hours of amassed acoustic data with the intent of applying learnings from the models generated for the audio data as an additional validation method for the camera trap data. Ultimately, the data we retrieve by applying the computer vision methods will help inform critical habitat areas within the A2A corridor and will identify areas for preservation or restoration.

Shannon Cressman

Hi, I’m Shannon Cressman, a lead biological science technician for the USFWS in Green Bay, Wisconsin. My job entails surveying angler caught salmonids that reside in Lake Michigan and Lake Huron of the Great Lakes. I have my B.S. in Marine Biology with a concentration in biological sciences from the University of Maine at Machias.

Project: Our program, the Great Lakes Fish Tagging and Recovery lab ages over 3,500 structures from both hatchery and wild-caught salmon and trout. Specifically, we analyze scales from Chinook Salmon and Steelhead, as well as otoliths from Lake Trout. These structures are aged within three months during the winter, leading to a significant workload. To address this, I aim to create a program using AI to detect annular rings in aging structures to estimate age. This will streamline the aging process and ensure consistency among agers across state, federal, and tribal agencies throughout the Great Lakes region.

Nina Ferrari

I'm a PhD student at Oregon State University working with Dr. Matt Betts and Dr. Erica Fleishman. My research aims to answer central questions of species coexistence theory and niche partitioning. I investigate the abiotic (microclimate) and biotic (competition and vegetation structure) drivers of vertical bird distributions in old-growth and second-growth forests. I utilize myriad techniques including tree climbing, machine learning of bioacoustics, and lidar to evaluate fine-resolution metrics across the vertical dimension in the H.J. Andrews Experimental Forest. I am particularly interested in how birds can mitigate climate driven changes by modifying their habitat use.

Project: The use of ARUs and bioacoustics is increasing in the field of avian ecology. However, most studies use data at coarse horizontal spatial resolutions to answer questions about population dynamics and site occupancy. I am interested in using bioacoustics data to answer questions about fine-resolution behavior and vertical habitat use of birds. I will build a multi-species classifier that can predict the height in a tree a vocalization occurs based on relative decibel levels across multiple ARUs. A goal this project is to better understand how different forest management practices (harvest regimes of second-growth stands, retention of old-growth trees) influence vertical habitat use by forest songbirds.

Katie Grabowski

I'm Katie, a first year PhD student at the University of British Columbia in Vancouver, Canada. For my PhD, I am working in Gorongosa National Park (GNP) in Mozambique, studying the effects of large carnivore reintroduction on the mesocarnivore community. Camera traps play a vital role in collecting information about these small, elusive nocturnal animals, and I am interested in leveraging the power of computer vision to help process the collected data more efficiently. Outside of my research, you can find me running, swimming, baking, or hanging out with my dog, Charlie.

Project: The widespread global loss of apex predators from ecosystems has led to dramatic changes in downstream trophic levels, including the release of smaller-bodied mesocarnivores. As apex predators are reintroduced to ecosystems where they were extirpated, it remains unknown how mesocarnivore populations will respond. For my project, I will use data from a camera trap grid in GNP that has been operating since 2016. My goal at CV4E is to train a model to identify the species and number of individuals for each image sequence, not only for mesocarnivores but for all large mammals, so that we may better monitor changes to the wildlife community in the wake of large carnivore reintroduction. My ecological questions will benefit from the use of computer vision to speed the timeline from the detection of an animal by a camera trap to generating usable data for building models.

Alexandra DiGiacomo

I am a PhD candidate in Biology at Stanford University. My dissertation research integrates remote sensing and biologging technology to investigate white shark movement ecology in Monterey Bay. Previously, I received a B.S. in Biology at Duke University and conducted research in Duke's Marine Robotics and Remote Sensing Laboratory. As a scientist, I am broadly motivated to seek solutions that expand the scale at which we can understand and conserve ocean ecosystems. For this reason, my research works to leverage and develop computational approaches to answer critical biological and ecological questions.

Mariam Ayad

Mariam is a PhD candidate at the University of California, Santa Cruz in the Ocean Science Department. She graduated from California State University of Los Angeles with an M.S. in Environmental Science with an emphasis in Geospatial Sciences and from the University of California Irvine with a B.S. in Earth System Science. Her master’s thesis work was on remote sensing and classification techniques to identify different types of marine pollution, specifically utilizing remote sensing and differentiating stormwater and wastewater runoff in the Tijuana River. This work was published to Frontiers in Environmental Sciences in December 2020 (Ayad et al., 2020). She received the NASA MSI fellowship to support her PhD research: “Detection and Analysis of Coral Stressors in Florida Keys and Belize Barrier Reef using Remote Sensing Imagery”. The goal of this project is to identify high-risk areas that are contributing to coral reef bleaching or inhibiting coral growth. She is very interested in coral reefs and their response to various chemical and physical changes in the ocean and climate as a whole. She hopes to incorporate remote sensing and field measurements to quantify the response of coral reefs to a changing climate.

Project: Coral reefs are vulnerable ecosystems threatened by anthropogenic stressors and climate change. About 40% of coral reefs have been lost in the past 40 years in the Caribbean Sea. Coral reefs face various stressors such as nutrient pollution, overfishing, sedimentation, increased sea surface temperatures (SST), and ocean acidification. I aim to develop a machine learning model with in-situ validation to identify healthy and bleached coral using PlanetScope SuperDove satellite imagery. This research would be of value to coral reef managers and stakeholders in terms of ocean sustainability projects, reef management practices, and long-term predictions of the ocean ecosystem.

Alice Michel

I am a PhD candidate at the University of California, Davis. I study intergroup interactions in western lowland gorillas. My research takes place in a community reserve in the peatland rainforest of northern Republic of Congo, where I use passive acoustic monitoring, genetics, and traditional field techniques to collect data alongside people from a local village. My hope is that the act of basic scientific research, aiming simply to ask questions about our world, will support better conservation outcomes in critical ecosystems. When I'm not searching for gorilla poop in the forest or gorilla sounds on the computer, I enjoy running, biking, and taking pictures.

Project: In some gorilla populations, adult male, or silverback, gorillas communicate with other groups by chest beating back and forth at night. With the end objective of facilitating my research on the form and function of this form of communication, the goal of this project is to develop computer vision methods to detect gorilla chest beats in five months of passive acoustic recordings. A subsequent goal is to determine if there is sufficient intraindividual repeatability and interindividual variability in chest beats with which to identify individual gorillas.

Sheila Holmes

I am a researcher at the department of Wildlife, Fish and Environmental Studies, Swedish University of Agricultural Sciences in Umeå, Sweden. My main research focus is measuring and understanding the impacts of human activities on various aspects of sustainable development. For example, I study how restoration activities in Madagascar’s humid forests affect biodiversity, human livelihoods, and zoonotic disease risk. I work closely with several academics and conservation NGOs, and I look forward to sharing the lessons I learn at the CV4E workshop with my students and collaborators, to support ongoing and future biodiversity monitoring efforts.

Project:Several conservation NGOs in Madagascar have ongoing reforestation programs but lack the resources to monitor the biodiversity outcomes of these programs. Though there is interest in using acoustic recorders to assist in monitoring, current analysis tools perform poorly at identifying rare tropical species. During the CV4E workshop, I will work on training a multi-species machine learning model to identify seed dispersing birds from AudioMoth recordings in forest, reforestation areas, and cleared/burned areas across the humid forest ecoregion. We can then use detections in occupancy modelling to identify factors that support biodiversity and seed dispersal function in reforestation areas.

Braden Charles DeMattei

I am the lab and data manager for Dr. Stephanie Hampton’s lab at Carnegie Science where we study the ecology of both rotifers and lakes under ice. I am interested in developing and applying computer vision and machine learning techniques to food web ecology and species interaction analysis. I received my B.S. in Marine Biology from UCLA and my MSc in Marine Conservation at the University of Aberdeen with a focus on advanced statistics. I have worked on everything from coral reef ecology to Scottish demersal mixed fisheries management to California Central Valley salmon population management.

Project: Applying computer vision to trophic interaction and food web ecology is something that has not yet been fully explored in the literature, as it is very difficult to collect enough data for most predator-prey pairs. Using the sessile rotifer, Rhinoglena sp., as the predator and the motile algae, Cryptomonas erosa, as prey, my project for CV4E will look at detecting when and how often a predator and its prey encounter each other. Rhinoglena remains stationary while it hunts and creates a feeding field with cilia to bring its prey to it. This makes it relatively easy to film the entirety of its hunting efforts without needing to refocus the microscope. Developing this algorithm will hopefully be the jumping off point for a grander project that will develop a predator-prey interaction detection and classification model that will allow for more in-depth analyses of freshwater food web ecology.

Guy Zer Eshel

I am Guy, I’m a PhD student with a deep interest in exploring the intersections of sensory ecology, movement behaviour, and neurobiology. My work extends encompassing diverse projects such as bioacoustics of moths, electrophysiological studies of ascidians, and sensory ecology investigations of whirling beetles. Alongside my research, I find immense fulfillment in teaching and mentoring.

Project: My current project focuses on the remarkable adaptations of whirling beetles (Coleoptera: Gyrinidae), which navigate the surface tension of freshwater ecosystems using dual visual systems and highly sensitive antennae. Early hypotheses suggest these beetles rely on returning surface waves for spatial orientation, but this has yet to be thoroughly tested with modern tools. By employing high-speed imaging and computer vision algorithms, my research aims to track these surface waves and analyse beetle behaviour in response to them.

Kalindi Fonda

I’ve been crafting my tech for nature journey over the last few years, sometimes more tech, sometimes more nature and sometimes more journey. My experience is a collage of projects, roles and learning that have given me valuable insights into the tech for nature space. I’ve worked on placing underwater cameras to monitor mantas, filmed stories for a documentary about a wildlife reserve in Namibia, and contributed to an ocean bioacoustics project as my 20% engagement while at Google. Currently, I am building Nature Venture, an organisation to accelerate nature projects. I also collaborate with Wildlife.ai and the New Zealand Department of Conservation on a marine reserve and fish monitoring via BUV (Baited Underwater Videos) data and ML project. I worked in ed-tech for a big part of the last decade, because of its mission of making education and knowledge accessible, and I am still involved by sharing and giving talks on the ways we can use tech for good especially thanks to its leverage and power to scale. I crossed the Atlantic Ocean in a sailboat at the beginning of last year, where I wrote most of my thesis The Role of Biodiversity in Business. I love learning. I cycle and play underwater rugby.

Project: Two years ago I spent some time at LAMAVE's Manta research station in Palawan, Philippines. We deployed underwater camera traps, which took a photo every five seconds of areas of interest, be it coral that might be a cleaning station or a feeding corridor. Every couple of days, we would replace the batteries, retrieve the SD cards and return with thousands of photos: 10.000 per camera per day. We would then go through these by hand to identify manta ray sightings and other species such as sharks, turtles, and rays. We would also recognize individual mantas based on the unique spot patterns on their bellies. Each step of this process was manual, and my CV4E project aims to make this process simpler and more automated: from reducing the amount of "empty" images (those containing nothing but sea and small fish, the images containing animals of interest represented a small portion), identifying species of interest, or even manta individuals. I am excited to see how far we'll get in the 3 weeks of the summer school! Streamlining this process would enable LAMAVE and similar organizations to access research insights more quickly and with fewer resources. This kind of work provides essential data for research (monitoring the health of marine ecosystems, and identifying critical habitats) and supports policy efforts, such as establishing and preserving marine reserves. Additionally, I believe that by showing examples of tech for good we can inspire more people to join this mission!

Matthias Zuerl

My name is Matthias, and I am a PhD student at the Machine Learning and Data Analytics Lab in Erlangen, Germany. I am particularly interested in how videos can serve as the basis for animal re-identification and behaviour recognition. A key research question that drives my work is whether an animal’s movement can be used as a unique feature to determine its identity. After completing my Master of Physics, I have worked on several projects applying machine learning to biological data, mainly for (polar) bears, and I value the collaboration with experts in the field. When I need a break from work, I make music, play ice hockey for the Nürnberg Bears (obviously), or cheer for the Nürnberg Ice Tigers.

Project: We are developing a framework designed to consolidate video-based re-identification methods for animals. Our goal is to integrate different backbone architectures in a modular way, allowing the most suitable model to be identified for each species. Currently, we are working with two datasets: one for polar bears and another for grizzly bears. The ultimate aim is to create a toolbox for the scientific community, enabling other research groups to apply these methods to additional species.