RFP/JOB: OpenCV Developer for Facial Recognition of Lions
This RFP looks for a developer or team that can aid in the creation of a Computer Vision engine
to identify lion individuals from frontal images.
Start Date: OCT 1, 2014
Timeline: 3 months
ABOUT US:
The Lion Guardians have been conserving lions and preserving cultures since 2007 with our mission to
promote sustainable coexistence between people and lions using cultural values, community
participation and science.
The Lion Guardians are a conservation organization that promotes the coexistence of people and lions
across Kenya and Tanzania. We have more than 70 East African team members protecting lions over
4,500 square kilometers (one million acres) of non-protected critical lion habitat. We have been able to
reduce lion killing by 99% and have documented a near tripling of the lion population at our core site
in the greater Amboseli ecosystem.
PROJECT OVERVIEW:
The Lion Guardians in conjunction with IEF R&D and partner conservation initiatives are developing a
lion database/website and computer vision ID system for east Africa. The goal is to address the
problems of fractured habitat and the marginalization of “sparsely populated’ wildlife areas. Recent
studies have shown greater mobility by lions through these areas, linking previously believed disparate
populations into new population boundaries.
These studies have been slowed though by a lack of cross pollination between data sets, prohibiting the
correlation of lion ID’s which encompass an individuals full range. Therefore this initiative aims to
unify this data with an easy to use application to identify and consolidate lion sightings in east Africa.
The computer vision component of this project will correlate ID matches from static images (non real
time) submitted to the database. The main CV engine will be developed using OpenCV (2.4.9) with the
end goal of code reuse for future community development.
Developer/s project goals:
The developer/s will be responsible for the code base and algorithms used in the computer vision ID
engine. The end result being a command line application that will use a reference image to compare for
matches across a series of selected images. The application will need to contain multiple stages of
image processing from lighting normalization to feature identification and template matching,with the
goal ID rate being 75-85% across an averaged data set of “on the ground” lion images.
The project will focus on frontal face images to help with standardization and field collection of the
images.
To allow for future development the code base will be structured using a clean, modular approach that
will allow for future collaboration and development. Incorporating clear structural design and
commenting of the code in the development process.
Key Deliverables:
• Developing in C/C++ create an efficient algorithm to processes, normalize and compare an
image within a field of templates to produce a 75-85% match rate for individual lions.
• Develop and maintain the CV engine code base using Git, Github and Doxygen, documenting
and clearly structuring the code for easy reuse and collaboration.
• Communicating with the development team to allow for easy integration into the larger
website/database.
• Minor visual output for testing, debugging and showcasing purposes of the “matching” process.
Most of the work will not require a visual component.
Skills Set Background:
• Strong C/C++ programming background with development for Linux based systems using the
GNU tool chain including Make, Cmake, GCC.
• Developing C++ applications on remote server, specifically CentOS development.
• Community code maintenance using Git and Github and Doxygen.
• A strong background in computer vision, OpenCV and facial recognition with a emphasis on
template matching, common vision algorithms Eigenfaces, Fisherfaces, Laplacianfaces and
neural networks such as Adaboost.
• Ability to design to the specific application and communicate trade offs in time/complexity to
functionality.
• Good person to work with, experience with lions a plus.
If interested please contact after Sept 17:
Justin Downs: Justin@iefrd.com