SLOOP:  Photographic Animal ID System

PEOPLE


Sai Ravela

PI MIT

Pattern Recognition

Image Processing


Joaquin Salas

Co-Investigator, CICATA Queretaro-IPN

Pattern Recognition

Image Processing


James Duyck

Researcher, MIT

Skinks, Rank Aggregation, Crowd Sourcing and Relevance Feedback

Indexing and Retrieval System


Chelsea Finn

Researcher, MIT

Whale Sharks, Scale-Cascaded Affine and Relevance Feedback



Collaborators


Jason Holmberg

Ecocean USA

Whale-Shark


Andy Hutcheon

Department of Conservation, New Zealand

Skinks, Geckos


Alumni


Dr. Lloyd Gamble,

Lead Researcher,

Marbled Salamander


Fernando Shao

Researcher, MIT

Image Segmentation

Indexing and Retrieval System


Chris Yang

Researcher, MIT

Image Segmentation

Indexing and Retrieval System


Josh Runge

Researcher, MIT

Illumination

Indexing and Retrieval System


Michael Newton

Researcher, MIT

SCA & Point Matching

Indexing and Retrieval System


Emily Pitts

Researcher, MIT

SCA & Point Matching

Indexing and Retrieval System

 

New Paper: SLOOP: A Pattern Retrieval Engine for Individual Animal Idenfication

Visit the new site: sloop.mit.edu

KEY ATTRIBUTES OF SLOOP approach to Vision-based Animal Biometrics (2004-present)

(These points are notable, especially as some recent approaches appear unaware.)

  1. 1. First to demonstrate use of generic visual features for animal identification.
  2. 2. Fully automated systems are not ready in our opinion, but inching closer. Some pre-processing remains necessary in practice. Sloop is built as a retrieval system with relevance feedback. Focus on rare and endangered species.
  3. 3. Interactive preprocessing for segmentation, rectification and illumination.
  4. 4. Generic visual features and strategies include appearance (patches, local features), geometry (scale-cascaded alignment).
  5. 5. First to incorporate relevance feedback from crowdsourcing and incremental online learning.

Species

Salamanders, Grand Skinks, Otago Skinks, Geckos, Whale Sharks and more! Visit the new site: sloop.mit.edu

  1. Marbled Salamander SLOOP-A(CS) v2.7: The first release for marbled salamanders. Over 10,000 images, includes relevance feedback and crowdsourcing experiments. Now RETIRED.
  2. Grand and Otago Skinks SLOOP-SK(CS) v3.0: The release for Grand and Otago Skinks. Over 20,000 images, operational, see news article here.
  3. Whale Shark Sandbox SLOOP-WS(CS) v3.1: Currently in prototyping.
  4. Generic Sandbox SLOOP-G(CS) v2.7: A sandbox for us to help you index.
  5. Others: Geckos, Toads and many others!


METHODS PAGES



PRESS

  1. Finding a gecko in the crowd
  2. Program will identify skinks