Action Detection for Wildlife Monitoring with Camera Traps Based on Segmentation with Filtering of Tracklets (SWIFT) and Mask-Guided Action Recognition (MAROON)

Action Detection for Wildlife Monitoring with Camera Traps Based on Segmentation with Filtering of Tracklets (SWIFT) and Mask-Guided Action Recognition (MAROON)

Behavioral analysis of animals in the wild plays an important role for ecological research and conservation and has been mostly performed by researchers. We introduce an action detection approach that automates this process by detecting animals and performing action recognition on the detected anima...

Full description

Bibliographic Details
Main Authors: Frank Schindler, Volker Steinhage, Suzanne T. S. van Beeck Calkoen, Marco Heurich
Format: Article
Language:English
Published: MDPI AG 2024-01-01
Series:Applied Sciences
Subjects:
wildlife monitoring
deep learning
video instance segmentation
mask-supported action recognition
triple-stream convolutional neural network
action detection for deer
Online Access:https://www.mdpi.com/2076-3417/14/2/514

Internet

https://www.mdpi.com/2076-3417/14/2/514

Similar Items