Reasoning About Neural Network Activations: An Application in Spatial Animal Behaviour from Camera Trap Classifications

2018 ◽

Vol 17 (04) ◽

pp. 1850021

Author(s):

Gyanendra K. Verma ◽

Pragya Gupta

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Detection System ◽

Camera Trap ◽

Wild Animal ◽

Deep Convolutional Neural Network ◽

Multimedia Data ◽

Machine Learning Techniques ◽

Wild Animals ◽

Animal Detection

Monitoring wild animals became easy due to camera trap network, a technique to explore wildlife using automatically triggered camera on the presence of wild animal and yields a large volume of multimedia data. Wild animal detection is a dynamic research field since the last several decades. In this paper, we propose a wild animal detection system to monitor wildlife and detect wild animals from highly cluttered natural images. The data acquired from the camera-trap network comprises of scenes that are highly cluttered that poses a challenge for detection of wild animals bringing about low recognition rates and high false discovery rates. To deal with the issue, we have utilized a camera trap database that provides candidate regions utilizing multilevel graph cut in the spatiotemporal area. The regions are utilized to make a validation stage that recognizes whether animals are present or not in a scene. These features from cluttered images are extracted using Deep Convolutional Neural Network (CNN). We have implemented the system using two prominent CNN models namely VGGNet and ResNet, on standard camera trap database. Finally, the CNN features fed to some of the best in class machine learning techniques for classification. Our outcomes demonstrate that our proposed system is superior compared to existing systems reported in the literature.

Download Full-text

Deep learning based evaluation of the nightly behaviour of African ungulates in zoos

10.22541/au.161053712.25635429/v1 ◽

2021 ◽

Author(s):

Max Hahn-Klimroth ◽

Tobias Kapetanopoulos ◽

Jennifer Gübert ◽

Paul W. Dierkes

Keyword(s):

Neural Network ◽

Deep Learning ◽

Convolutional Neural Network ◽

Model Averaging ◽

Time Frame ◽

Animal Behaviour ◽

Video Data ◽

Post Processing ◽

African Ungulates ◽

High Degree

1. The description and analysis of animal behaviour over long periods of time is one of the most important challenges in ecology. However, most of these studies are limited due to the time and cost required by human observers. The collection of data via video recordings allows observation periods to be extended. However, their evaluation by human observers is very time-consuming. Progress in automated evaluation, using suitable deep learning methods, seems to be a forwardlooking approach to analyse even large amounts of video data in an adequate time frame. 2. In this study we present amulti-step convolutional neural network system for detecting animal behaviour states, which works with high accuracy. An important aspect of our approach is the introduction of model averaging and post-processing rules to make the system robust to outliers. 3. Our trained system achieves an in-domain classification accuracy of >0.92, which is improved to >0.96 by a postprocessing step. In addition, the whole system performs even well in an out-of-domain classification task with two unknown types, achieving an average accuracy of 0.93. We provide our system at https://github.com/Klimroth/Video-Action-Classifier-for-African-Ungulates-in-Zoos/tree/main/mrcnn_based so that interested users can train their own models to classify images and conduct behavioural studies of wildlife. 4. The use of a multi-step convolutional neural network for fast and accurate classification of wildlife behaviour facilitates the evaluation of large amounts of image data in ecological studies and reduces the effort of manual analysis of images to a high degree. Our system also shows that post-processing rules are a suitable way to make species-specific adjustments and substantially increase the accuracy of the description of single behavioural phases (number, duration). The results in the out-of-domain classification strongly suggest that our system is robust and achieves a high degree of accuracy even for new species, so that other settings (e.g. field studies) can be considered.

Download Full-text