DenseNet201 for Animal detection and repellent system
M Naveenkumar, Rakshitha A. Manoj, B Nandhakumar, Rushikesh Rahul · 2022
Animals entering the agricultural land near the forest areas destroy crops or even attack people residing nearby villages. For humans, agriculture crops, and animals to survive it is essential to utilize technology in a useful way. Therefore, there is a need for a system which detects the presence of animals and produces sound to repel the animals. Our study develops a model utilizing deep convolutional neural networks that detects animals through video analytics. The different features like color, Gabor and LBP are extracted using the segmented images of the animal. Possibilities of combining these features for improving the performance of the detection and classification have also been explored. Detection of animals is accomplished using CNN and symbolic classifiers. For validating the performance of the proposed algorithmic models and also due to non-availability of a large benchmarking related dataset, successful attempts to create an animal image dataset and an animal video dataset. Experimental results show that better detection accuracy is obtained for even partial images of animals. The system will give better accuracy in terms of f1 score precision, recall and it will support the society to safeguard agriculture, farmland, wildlife and human beings.