Quick Links
This repository holds the source files used to generate my 2021 Ph.D. Dissertation titled, "Animal Detection for Photographic Censusing." This dissertation is a continuation of my 2015 master's thesis, "Photographic Censusing of Zebra and Giraffe in the Nairobi National Park".
This dissertation document is meant to partially fulfill the requirements for a Ph.D. degree in Computer Science at RPI. The thesis is paired with an oral defense, given in November 2021. The final draft was submitted to the University in November 2021 for graduation in December 2021.
Animal population monitoring is hard to do at large scales. It is too logistically demanding to track thousands of animals with invasive tools like ear tagging, and methods like aerial surveys and hand-based counts cannot track individuals over time. A database of unique animals and their sightings can be a critical tool for conservation; ecologists gain a more intimate and timely understanding of an endangered species' health when they can estimate life expectancy, visualize migration patterns, and quickly measure the effects of conservation policies.
This dissertation proposes photographic censusing, a way to visually track the population of an entire species with as little human effort as possible. The method is based on a two-day event called a photographic censusing rally, formed as a sight-resight study (building off of mark-recapture) to estimate the size of the population. Photographic censusing is highly automated, is designed to be bootstrapable for new species, and uses citizen scientists to collect large volumes of photographs across a large geographic area. A novel 5-component animal detection pipeline is proposed to analyze collected images of animals and filter sightings of animals for ID. The pipeline offers a whole-image classifier for quick filtering, a bounding box localizer to find annotations, an annotation labeler to determine species and viewpoints, a coarse segmentation algorithm to mask the background, and a component to recognize poor sightings, and is evaluated on new datasets.
This research also presents the Great Grevy's Rally (GGR) results from 2016 and 2018. These censusing events attempted to catalog the entire resident population of Grevy's zebra (Equus grevyi) in Kenya and, combined, collected over 90,000 images from more than 350 volunteers. The GGR analysis in 2018 was done with automated tools but still required large amounts of work (~18,500 human decisions), cost USD $50,000+, and took over three months. This dissertation discusses the work needed during a photographic census and analyzes failure modes that require human interaction. The novel concept of Census Annotation (CA) is introduced to find comparable regions of animals for automated ID, which drastically increases automation. The 56,588 images from GGR 2018 were reprocessed with the latest recommended methods presented in this work; 11,916 annotations were automatically found for comparable, right-side Grevy's zebra; ID curation used 1,297 human decisions before converging, and 2,820+/-167 Grevy's zebra were estimated to be in Kenya in 2018. This result is consistent (within 0.3% of the original estimate of 2,812+/-171) with previous estimates on GGR 2018 data and was achieved with a 93% reduction in human effort.
- Dr. Charles Stewart, Advisor, Computer Science Department Chair
- Dr. Barbara Cutler
- Dr. Bülent Yener
- Dr. Richard Radke (Department of Electrical, Computer, and Systems Engineering)
- Dr. Tanya Berger-Wolf (Ohio State University)
- List of Tables
- List of Figures
- Acknowledgement
- Abstract
- Chapter 1 - Introduction
- Chapter 2 - Related Work
- Chapter 3 - Animal Detection Pipeline
- Chapter 4 - Overview of Photographic Censusing
- Chapter 5 - Census Annotation
- Chapter 6 - Photographic Censusing of Grevy's zebra in Kenya
- Chapter 7 - Conclusion
- References
- Appendix A - GGR 2018 Participant Guide
- Appendix B - Chapter attributions & Copyright Permissions
The public portion of the Ph.D. defense may be watched in the file defense.mp4 (runtime 61 minutes), and the slides are available as defense.pdf.
Alternatively, the defense presentation can be watched on YouTube.
@phdthesis{parham_animal_detection_2021,
title = {Animal Detection for Photographic Censusing},
author = {Parham, Jason R.},
year = 2021,
address = {Troy, NY, USA},
school = {Deptartment of Computer Science, Rensselaer Polytechnic Institute},
type = {Ph.D. Dissertation}
}This dissertation reproduces previously published works and copyrighted figures. The dissertation attributes the papers and their copyright holders in the chapters where they are used for body text, tables, and figures. Original PDF copies of each paper, along with their copyright permissions, can be viewed in the attribution folder. In addition, appendix B of the dissertation document gives a breakdown of where each paper's content is used.
- J. Parham and C. Stewart, "Detecting plains and Grevy’s zebras in the real world," in IEEE Winter Conf. Applicat. Comput. Vis. Workshops, Lake Placid, NY, USA, Mar. 2016, pp. 1–9.
- J. Parham, et al., "An animal detection pipeline for identification," in IEEE Winter Conf. Applicat. Comput. Vis., Lake Tahoe, CA, USA, Mar. 2018, pp. 1–9.
- J. Parham, J. Crall, C. Stewart, T. Berger-Wolf, and D. I. Rubenstein, "Animal population censusing at scale with citizen science and photographic identification," in AAAI Spring Symp., Palo Alto, CA, USA, Jan.2017, pp. 37–44.
- J. Parham, C. Stewart, T. Berger-Wolf, D. Rubenstein, and J. Holmberg, "The Great Grevy’s Rally: A review on procedure," in AI Wildlife Conserv. Workshop, Stockholm, Sweden, Jul. 2018, pp.1–3.
make all| Course | Credits |
|---|---|
| Computer Operating Systems | 3 |
| Cryptography & Network Security I | 3 |
| Randomized Algorithms | 3 |
| Cryptography & Network Security II | 3 |
| Machine Learning | 3 |
| Programming Languages | 3 |
| Computational Vision | 3 |
| Neural Networks for Computer Vision | 3 |
| Computer Vision for Visual Effects | 3 |
| IBEIS Software Architecture for Algorithm Development | 6 |
| IBEIS Software Services for Cloud Computing | 3 |
| Required Dissertation (Dr. Yener) | 9 |
| Required Dissertation (Dr. Stewart) | 27 |
| Additional Dissertation (Dr. Stewart) | 63 |
| TOTAL | 135 |
This dissertation is a product of patience. The highest amounts of mental and physical endurance have been given by my wife, Lindsay. Her skill in raising our children, Heidi and Lincoln, has been awe-inspiring, and I thank her sincerely for her perseverance. It can be challenging to explain to young children why work can be so important -- important enough to miss dinners or playtimes at the park. I believe and hope that someday they will understand that the missed time together was instead invested into a higher, more urgent obligation. I dedicate this work to them, as they will be the truest beneficiaries of any success my work may find in the pursuit of wildlife conservation.
I also thank my advisor, Dr. Charles Stewart, who has been a compassionate guide in my academic career and in life as a young husband and father. I appreciate his patience and expertise and the guidance from my committee. The generosity and flexibility of my employers during this Ph.D. process is something that I'm not sure I completely understand; I thank Drs. Anthony Hoogs, Matt Turek, Rusty Blue, and Keith Fieldhouse at Kitware, along with Jason Holmberg and Dr. Tanya Berger-Wolf with Wild Me. I also thank the Gordon and Betty Moore Foundation for their financial support. My graduate lab partners at RPI, Drs. Jon Crall and Hendrik Weideman, provided excellent discussions and stimulation on the latest machine learning methods, and I thank Dr. Barbara Cutler for her tranquility in indulging our energy. Jon, if you ever want to camp in the African bush, just let me know. I also thank my peers in machine learning for animal conservation, Sara Beery and Dr. Stephan Schneider, for their work in kindling a small but passionate research community. Drs. Dan Rubenstein, Kaia Tombak, and Megan McSherry have also been instrumental in facilitating this research, and I thank them for their diligence in working with me over the years. I also cannot forget the dedication and benevolence of the research staff at the Ol Pejeta and Lewa conservancies, the Great Grevy's Trust, the Kenya Wildlife Service, and numerous Wildbook projects.
Lastly, I would like to thank my parents, Anthony, Grace, Linda, Harlon Jr., Kent, and Julie, and my siblings Stephany, Harlon III, Joyce, Kelsey, Kyle, and Chad for their continued support. I also appreciate my co-workers at Wild Me, Jon Van Oast, Drew Blount, Colin Kingen, Mark Fisher, Ben Schiener, and Tanya Stere for permitting my chaos and giving me a fulfilling place to work with friends. I also thank Drew and Olga Moskvyak for their work on new detection components and Tanya as honorary editor. Specific thanks to my sisters-in-law Brittany and Kelsey Sundman, and to Ben and Kaia, for their last-minute help, looking at some zebras when nobody else really wanted to.
- animal detection
- animal censusing
- census annotation
- census annotation region
- photographic censusing
- photographic censusing rally
- citizen science
- computer vision
- machine learning
- Kenya
- IBEIS
- Wildbook
- Wildbook IA
- WBIA
- Kenya Wildlife Service
- Grevy's zebra
- Great Zebra & Giraffe Count
- Great Grevy's Rally
- Great Grevy's Rally 2016
- Great Grevy's Rally 2018
- GGR
- http://www.greatgrevysrally.com
- https://github.com/WildbookOrg/wildbook-ia
- https://pypi.org/project/wildbook-ia/
- https://registry.hub.docker.com/r/wildme/wildbook-ia
- https://github.com/Erotemic/crall-thesis-2017
- https://hjweide.github.io/research/
- https://olgamoskvyak.github.io
- https://github.com/drewblount
- https://wildme.org