Animal Intrusion Detection Using Deep Learning for Agricultural Fields

Abstract

Crop yield is severely threatened by animal invasion, which has consequences for both food security and farmer income. This suggested model details how this issue might be addressed via the integration of IoT and ML methods. As image processing and Internet of Things (IoT) sensor monitoring network technologies have advanced, sensors have undergone radical changes. Animal-human conflict is a serious issue in the agricultural sector and the forest zone, posing a threat to human life and causing significant material loss. Wireless sensors can analyze video clips from a gathered dataset to create an animal incursion detection system. This study introduces Invariant Feature Extraction (FE) using SlowFast architecture. The videos are annotated first, and then the IFE model is used to extract the spatial data. Classifying animals based on their pictures, can keep better tabs on them. Accidents involving animals and vehicles may be avoided, animals can be traced, and theft can be avoided if detection and categorization methods are used. The efficient deep learning techniques are helpful for this.