BIRDS RECOGNITION USING DEEP LEARNING BASED ResNet-50 MODEL WITH SVM CLASSIFIER

Abstract Due to anomalous changes and for the management of the environment, monitoring and surveillance of animal species is a crucial duty. Because increasing environmental change puts many animal species in danger, a detailed understanding of the intricacy of the natural environment would be preferable. Installing technological systems with the knowledge to address the issues mentioned earlier is necessary for the best protection of wild animals and the best monitoring of livestock. This article offers the ResNet50 model, built on deep learning and trained using SVMs as a classifier. This work combines the SVM classifier with several deep-learning models, including VGG_16, VGG_19, ResnNet101, DensNet121, and Mobil Net. Similar to the previous approach, this one uses support vector machines to classify birds after extracting their characteristics. The SoftMax layer is swapped out for SVM in the suggested techniques. Using images of birds, the performance of the employed models is evaluated. Support vector machines are utilized to categorize the appearances of birds in images after a ResNet50 architecture is developed to simulate their looks. Instead of using the SoftMax approach, this technique supports vector machines for a more reliable and accurate categorization of bird images. Then, another deep learning approach was created based on support vector machines, including VGG_16, VGG_19, ResnNet101, DensNet121, and Mobil Net. The suggested approaches offer improved accuracy compared to conventional methods because support vector machines are more potent classifiers than SoftMax. The performance results demonstrate that the suggested models perform better than the standard deep learning, such as VGG_16, VGG_19, and ResNet101, DensNet121, and Mobil Net-based SVM models.