Hybrid Deep Neural Network for Alpaca Animal Recognition: Achieving High Accuracy in Detection and Classification

Developing a reliable and efficient model of animal recognition is challenging. There are numerous factors to consider, such as the model's accuracy, the number of variables, the complexity of the calculations, etc. We propose the Hybrid Deep Neural Network (HDNN) to make the technology more applicable to detect and a set of classifiers to classify. This new HDNN attempts to establish a balance between detection accuracy and classifying. It comprises two modules-feature extraction and classification modules-that collaborate to achieve its objectives. Yolo v5 comprises the feature extraction module's hybrid deep neural network architecture. Here, two neural networks are employed: one for extracting image characteristics and the other for identifying objects within an image. The classification component employs two completely interconnected layers as a classifier. It achieves an outstanding balance between recognition accuracy and model size compared to other methods. The proposed method has been shown to be effective in experiments, with an accuracy of 98.02% in classification and high precision values. The proposed method can automatically categorize animals while accounting for various dataset difficulties.