Identification of Animals Through Footprint Analysis

The study aims to develop an efficient and accurate system for automatically detecting and classifying animal footprints in various environmental conditions. Leveraging the capabilities of YOLOv8, the proposed method demonstrates robustness in identifying diverse animal species based on their footprints. The research involves dataset curation, model training, and performance evaluation to validate the effectiveness of the developed system. Results indicate promising accuracy rates, showcasing the potential of YOLOv8 for wildlife monitoring and conservation efforts. This work contributes to the advancement of computer vision techniques in the field of ecological research and animal behaviour studies. We analysed a dataset comprising 1567 images distributed across 10 classes. Here we propose an animal identification system that employs image processing techniques. Firstly, the collected footprint images are pre-processed. Images are converted into Gray-scale then many filters are applied for accuracy like saturation, exposure and labelling images in the dataset are also done in a proper way. The thorough testing and validation processes used in the project are essential to proving the categorization model's dependability and correctness. The assessment covers a wide range of datasets, including ones with hazy photos, to reflect the complexity of actual animal detection situations. To summarise, the methods and unique characteristics of the Animal Detection using Footprints project are explained in this abstract. In addition to advancing animal identification systems, the study emphasizes the value of uncertainty concerns in practical applications by departing from conventional deblurring techniques and adopting a probabilistic model.