Footprint-Based Animal Identification: A Hybrid Model Combining CNN, SVM, and Transfer Learning for Wildlife Encroachment

Animals can be identified by using various machine learning techniques among which a hybrid model combines multiple techniques together. In this approach a hybrid model with Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), and Transfer Learning is used to recognize and classify animal footprints. The goal is to prevent animal intrusion into human dwellings and protect endangered species by enabling timely detection. Using a pre-trained CNN (e.g., VGG16) for feature extraction, the system identifies distinct footprint patterns from images. These features are then passed to an SVM for accurate classification. The model benefits from Transfer Learning, leveraging knowledge from large-scale datasets to enhance detection accuracy, even with limited footprint data. This approach enables efficient monitoring of wildlife near human settlements, allowing for proactive measures to prevent conflicts and supports biodiversity protection.