Prediction of Human Animal Conflict Through Machine Learning Based HAR Algorithm

View at publisher Summarise with VIRAT
Authors: D Preethi, Valarmathi R S, Bhuvaneshwari. V, V. Prithivirajan, V Gnanaprakash, Vaishnavi P
Cited by: 0
References: 11

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Conflicts and interactions that arise when human wants and activities directly or indirectly collide with those of wild animals are referred to as human-animal conflicts. This conflict can occur in a variety of locations, including rural, urban, and natural settings, and it can take many different forms. Crop raiding, livestock predation, human attacks, and disease transfer are a few frequent interactions with animals. Therefore, it is crucial to keep an eye on animal behaviour and movement patterns, undertake study on it, and be aware of better management techniques and early conflict detection systems. Machine learning algorithms are used to observe and prevent the hunting of endangered species for purposes other than self-defence because it has been seen that human attacks on these animals are growing. The human action recognition (HAR) refers to the comprehension of actions made by a human in a video. Actions modelling is mapping an action to a label that clearly defines a particular occurrence of that activity from various angles and at various speeds. In addition, there are a number of difficulties with newline completely automated HAR systems, including background clutter, newline occlusion, scale, and look. Cost-effective equipment makes it simpler to collect 3D data. As a result, numerous objects and people in a scene can provide depth information in the form of depth pictures. An updated learning method called Self-newline adaptable Differential Evolution-Extreme Machine Learning (SaD-EELM) for action classification is used to overcome problems and challenges of HAR, such as background clutter. Single Layer Feed- forward Network (SLFN) with Depth Motion Maps-Local Newline Binary Pattern (DMM-LBP) is used to improve action classification utilizing depth-based input. Furthermore, newline misclassification is decreased by using the Kernel-based Extreme Learning Machine (KELM) classifier.

Fields of Study & Subjects

Human Pose and Action Recognition Wildlife Ecology and Conservation Smart Agriculture and AI

Publication details

Journal / source
International Conference on Intelligent Data Communication Technologies and Internet of Things (IDCIoT)
Pages
128–133
Publication date
4 February 2026
Language
EN