Computer Vision Methods for Ecology

Course Content

Computer vision (CV) is significantly accelerating ecology research by automating the analysis of raw images and video from camera traps, drones, and satellites. While ecologists often have training in statistics and programming, they are rarely exposed to the software engineering, machine learning and CV skills necessary to analyze big sets of visual data on their own. The Workshop on Computer Vision Methods for Ecology (CV4E), is an intensive 3-week workshop designed to fill this educational gap. Our mission is to empower ecologists to efficiently process their existing data, design new studies around CV, and scale their research to larger datasets.

The workshop is built on four key pillars: (1) Teach applied CV as a tool for ecological research by providing instruction and tools specifically designed for ecologists, built around real ecological data. (2) Empower ecologists to build their own CV-based systems by building intuition, skills and confidence by having them work with their own data to address questions about the natural world. (3) Grow the CV-for-Ecology community, a hub where expert ecologists and leading CV scientists can exchange ideas and best practices to address real-world challenges in conservation and sustainability. (4) Provide necessary computational resources to store data, collect annotations, train models, and host solutions during the school.

CV4E a three-week full-immersion course, composed of classroom training and hands-on projects with one-on-one mentorship. The students will be taught the rudiments of computer vision and how to train and evaluate computer vision models on their own data to help answer specific ecological research questions. Students will leave with a working tool, a grasp of the underlying concepts, and the ability to tackle diverse ecological problems with computer vision. School participants will also develop a network of computer vision researchers with whom they can engage and collaborate. Participants can expect access to GPU-accelerate cloud compute for the duration of the course.

Course Format

The format is three-weeks in person geared to early career scientists and graduate students. Participants will arrive with data they have collected to address an ecological research question. Leading up to the class, instructors will help students get their data prepared for the CV techniques they will experiment with. Students should anticipate dedicating most of each day to the class. They will listen to lectures, hear from guest speakers, participate in group activities, and have dedicated work sessions in small groups with a dedicated instructor. 

What’s Included

The total cost of the program includes:

• Extensive pre-course interactions with instructor team to prepare for the course
• A shared room at the SMSC Residence Hall (single rooms available for extra cost)
• Full meal package including 3 meals/day and 2 coffee/tea breaks (Mon-Fri only) at the SMSC Dining Commons
• Access to remote GPU-accelerate compute and storage
• In-person lectures and group activities
• Small group work with dedicated instructors focused on thematically similar projects
• Optional weekend social outings

Software and Programs

All work will be conducted in Python, the lingua franca of modern machine learning. Students will learn to use the VSCode integrated development environment to access remote compute, train CV models, and evaluate results. We expect students to learn the rudiments of Python before arriving on-site. Instructors are available to help students in that process and we have curated a list of online resources to assist. Students are also encouraged to seek out intro Python courses at the home institutions where available. 

Applying to this Course

As with all our other programs, you’ll need to create a profile, then complete an application to this course through our application site HERE. However, there are some additional steps required of those looking to complete an application to this course. Applicants will need to complete all the following steps to be considered:

1. Create a profile and complete and application on our application website. You’ll need the following materials ready when applying on that site:
   • Your CV. This will be uploaded when you create your profile in the system.
   • A “statement of interest”. This will be uploaded with your application. For this course, the statement should include the following 2 components uploaded as a single PDF document:
     1. 1. A 1-page project proposal (11pt font, 1″ margins) addressing the following:
           • Research problem — What question are you interested in and how would computer vision methods better enable you to address it?
           • Data — What type of data do you plan to work with? Do you have the data already in hand? How much data do you have, how much will you have by the time the summer school will start? Is the data labeled already, or do you need to develop a labeling plan?
           • Impact/Outlook — What is the likely impact of your research for science, policy, education and conservation?
        2. A 1-page Personal Statement (11pt font, 1″ margins) describing your accomplishments, skills and career objectives.

   • A link to a github repository with a programming example. This can be in Python or R. You’ll be asked to provide this link toward the end of the application.

2. Provide a description of your data using this form: https://forms.gle/H8kGs3xrB2dD3V6Q7

3. Have a professional letter of reference sent directly to [email protected] by your referee before the application deadline.

Email [email protected] with any questions.