loading page

An approach to rapid processing of camera trap images with minimal human input
  • +7
  • Matthew Duggan,
  • Melissa Groleau,
  • Ethan Shealy,
  • Lillian Self,
  • Taylor Utter,
  • Matthew Waller,
  • Bryan Hall,
  • Chris Stone,
  • Layne Anderson,
  • Timothy Mousseau
Matthew Duggan
University of South Carolina at Columbia
Author Profile
Melissa Groleau
University of South Carolina at Columbia
Author Profile
Ethan Shealy
University of South Carolina at Columbia
Author Profile
Lillian Self
University of South Carolina at Columbia
Author Profile
Taylor Utter
University of South Carolina at Columbia
Author Profile
Matthew Waller
University of South Carolina at Columbia
Author Profile
Bryan Hall
South Carolina Army National Guard
Author Profile
Chris Stone
South Carolina Army National Guard
Author Profile
Layne Anderson
South Carolina Army National Guard
Author Profile
Timothy Mousseau
University of South Carolina at Columbia
Author Profile

Abstract

Point 1: Camera traps have become an extensively utilized tool in ecological research, but the processing of images created by a network of camera traps rapidly becomes an overwhelming task, even for small networks. Point 2: We used transfer training to create convolutional neural network (CNN) models for identification and classification. By utilizing a small dataset with less than 10,000 labeled images the model was able to distinguish between species and remove false triggers. Point 3: We trained the model to detect 17 object classes with individual species identification, reaching an accuracy of 92%. Previous studies have suggested the need for thousands of images of each object class to reach results comparable to those achieved by human observers; however, we show that such accuracy can be achieved with fewer images. Point 4: Additionally, we suggest several alternative metrics common to computer science studies to accurately evaluate the performance of such camera trap image processing models, as well as methods to adapt the model building process to two targeted purposes.

Peer review status:UNDER REVIEW

16 Mar 2021Submitted to Ecology and Evolution
17 Mar 2021Submission Checks Completed
17 Mar 2021Assigned to Editor
18 Mar 2021Review(s) Completed, Editorial Evaluation Pending
23 Mar 2021Editorial Decision: Revise Minor
07 Apr 20211st Revision Received
08 Apr 2021Submission Checks Completed
08 Apr 2021Assigned to Editor
08 Apr 2021Review(s) Completed, Editorial Evaluation Pending
11 May 2021Editorial Decision: Revise Minor
16 Jun 20212nd Revision Received
18 Jun 2021Assigned to Editor
18 Jun 2021Submission Checks Completed
18 Jun 2021Review(s) Completed, Editorial Evaluation Pending